Contribution of Nonohnologous Duplicated Genes to High

Habitat Variability in Mammals

Satoshi C. Tamate,1 Masakado Kawata,*,1 and Takashi Makino*,1
1
Department of Ecology and Evolutionary Biology, Graduate School of Life Sciences, Tohoku University, Aoba-ku, Sendai, Japan
*Corresponding author: E-mail: kawata@m.tohoku.ac.jp; tamakino@m.tohoku.ac.jp.

Abstract
The mechanism by which genetic systems affect environmental adaptation is a focus of considerable attention in the
fields of ecology, evolution, and conservation. However, the genomic characteristics that constrain adaptive evolution
have remained unknown. A recent study showed that the proportion of duplicated genes in whole Drosophila genomes
correlated with environmental variability within habitat, but it remains unclear whether the correlation is observed even
in vertebrates whose genomes including a large number of duplicated genes generated by whole-genome duplication
(WGD). Here, we focus on fully sequenced mammalian genomes that experienced WGD in early vertebrate lineages and
show that the proportion of small-scale duplication (SSD) genes in the genome, but not that of WGD genes, is signif-
icantly correlated with habitat variability. Moreover, species with low habitat variability have a higher proportion of lost
duplicated genes, particularly SSD genes, than those with high habitat variability. These results indicate that species that
inhabit variable environments may maintain more SSD genes in their genomes and suggest that SSD genes are important
for adapting to novel environments and surviving environmental changes. These insights may be applied to predicting
invasive and endangered species.
Key words: habitat diversity, evolvability, adaptation, gene duplication, whole-genome duplication.

Introduction duplicated genes are immediately lost under these relaxed

The areas of habitats and environmental variability within
habitats considerably vary with species. Some species have
large habitat ranges that include various climatic conditions
and vegetation types, whereas other species have narrow
ranges that include one type of environmental condition.
Although geographical and ecological factors affect the
range limits, evolutionary factors affecting the capacity for
adaptation to novel and different environments are impor-
tant influences on range size and environmental variability
constraints, functionally redundant duplicated genes can be
maintained for a long time (Ohno 1970; Dean et al. 2008),
resulting in the accumulation of genetic variation. Genetic
variation in duplicated genes within a population is likely to
be maintained by their buffering effect (Wilkins 1997;
Hartman et al. 2001). Furthermore, a theoretical study
showed that duplicated genes are likely to be maintained in
gene regulatory networks in randomly fluctuating environ-
ments (Tsuda and Kawata 2010). Maintained duplicated
genes differentiate their function and/or expression pattern.

within habitats (Kirkpatrick and Barton 1997). For the genetic
factors, both a lack and an excess of genetic variation can
restrict their expansion to novel environments within habitat
boundaries (Hoffmann and Blows 1994). A recent study re-
ported that a Drosophila population with low genetic varia-
tion had low cold and desiccation tolerance and narrow
habitat distribution compared with a population with high
genetic variation (Kellermann et al. 2009). These observations
indicate that lack of genetic variation is related to failure in
adapting to novel environments.
Makino and Kawata (2012) focused on gene duplication as
the source of genome-wide genetic variation for adapting
various environments. Using 11 fully sequenced Drosophila
species, the authors showed that species with high proportion
of duplicated genes (PD) in a genome inhabit locations with
high environmental variability and proposed that PD in a
genome is correlated with adaptation to various environ-
ments in Drosophila. Gene duplication, particularly
common in eukaryotes, generates redundant gene copies,
and mutations are likely to accumulate in duplicated genes
under relaxed functional constraints. Although most
Article
Thus, gene duplication is a major source of genetic variation
in a genome. Although a positive correlation between PD and
habitat diversity was observed in Drosophila species, it is still
unclear whether the tendency holds true in other species.
Some groups of eukaryotes including vertebrates, plants,
and fungi have experienced whole-genome duplication
(WGD), and accordingly carry two types of duplicated
genes, one generated by WGD and the other by small-scale
duplication (SSD). Ohnologs, which are duplicated genes gen-
erated by WGD (Wolfe 2000), are not just ancient duplicated
remnants and have different features compared with SSD
genes (Maere et al. 2005; Blomme et al. 2006; Hakes et al.
2007; Wapinski et al. 2007). Ohnologs are likely to be dosage-
balanced genes, and thus rarely experienced gene duplica-
tions and losses during evolution (Veitia 2002, 2003, 2004;
Birchler et al. 2005; Makino and McLysaght 2010; Birchler
and Veitia 2012). Gene knockout of an ohnolog causes
lethal phenotypes compared with that of SSD genes as old
as ohnologs (Makino et al. 2009). Therefore, ohnologs are
unlikely to display not only copy number variations (CNVs)
of themselves in a human population (Makino and

The Author 2014. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please
e-mail: journals.permissions@oup.com

Mol. Biol. Evol. 31(7):17791786 doi:10.1093/molbev/msu128 Advance Access publication April 8, 2014 1779
Tamate et al. . doi:10.1093/molbev/msu128
McLysaght 2010) but also CNVs of their neighbor genomic
regions (Makino et al. 2013). Ohnologs are also associated
with dosage-sensitive functions, such as development,
transcription regulation, and protein complex formation
(Maere et al. 2005; Blomme et al. 2006; Brunet et al. 2006;
Hufton et al. 2008; Makino et al. 2009). SSD genes, in contrast,
are possibly related to response to biotic stimuli, a property
that could be important for adaptive traits (Blomme et al.
2006). In addition, Satake et al. (2012) showed that tissue-
specific expression of SSD genes was overrepresented in endo-
derm tissues (stomach, colon, liver, etc.) directly facing the
outer environment, whereas tissue-specific ohnologs were
likely to be expressed in ectoderm tissues (nerves system,
brain, eye, etc.) not exposed to the outer environment. We
accordingly hypothesized that SSD genes contribute more to
high environmental adaptability than ohnologs.
Drosophila do not experience WGD, so the effects of the
different types of duplicated genes on environmental adapt-
ability remain unknown. Mammalian species, in contrast,
experienced WGD twice in an early vertebrate lineage
(McLysaght et al. 2002; Dehal and Boore 2005; Nakatani
et al. 2007). The purpose of this study was to investigate
whether PD correlates with environmental variability even
in mammals and to test our hypothesis that the proportion
of SSD genes has a larger effect on the environmental variabil-
ity than the proportion of ohnologs. Using 30 fully sequenced
mammalian species, we performed comparative genomic
analyses to investigate the relationship between duplicated
geneseither ohnologs or SSD genesand environmental
variability.
Results and Discussion
PD Associated with Habitat Diversity
A previous research reported a relationship between PD and
habitat diversity for closely related species in the same genus
A B
1500
1000
Climatic envelope
Climatic envelope
500
0
-500
-0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15
Contrasts in proportion of duplicated genes
FIG. 1. Correlation between PD and the climatic envelope for euarchontoglires including Primates, Rodentia, and Lagomorpha. (A) Relationship between
PD and the climatic envelope. The x-axis represents PICs in the PD for 16 euarchontoglires. The y-axis represents PICs in the climatic envelope estimated
from WORLDCLIM data sets. (B) Relationship between proportions of lost duplicated genes and the climatic envelope. The x-axis represents PICs in the
proportion of lost duplicated genes for 16 euarchontoglire species. The y-axis represents PICs in the climatic envelope estimated from WORLDCLIM
data sets.
1780
MBE
(Makino and Kawata 2012). The fully sequenced mammalian
species used in our study were classified in only one genus per
order except for euarchontoglires including Primates
(nine species), Rodentia (five species), and Lagomorpha
(two species). Thus, we used 16 species of euarchontoglires
to investigate the relationship between genomic architecture
and habitat variability, employing a linear model in which
genome size, genome coverage and the total number of
genes reported by Ensembl (Flicek et al. 2011), the number
of duplicated genes, and PD were used as explanatory vari-
ables, and the climatic envelope (see Materials and Methods)
was used as a response variable. As a result of the model
selection, only PD was selected as an explanatory variable.
When the effects of the climatic envelope, habitat area, and
diet breadth on PD were examined, only the climatic envelope
was selected as an explanatory variable. When we investigated
the relationship between PD and the climatic envelope, there
was a significant positive relationship of PD with the climatic
envelope (R2 = 0.32, P = 0.012; fig. 1A and supplementary table
S1, Supplementary Material online) although there was no
correlation between PD and the climatic envelope over all 30
mammals (supplementary fig. S1, Supplementary Material
online). This result indicates that PD correlates with the cli-
matic envelope for closely related species (euarchontoglires in
fig. 1A and Drosophila [Makino and Kawata 2012]), but the
relationship was disrupted during evolution resulting in no
correlation for distantly related species (mammals overall).
It has been reported that duplicated genes consist of many
more recently duplicated genes than old duplicates (Lynch
and Conery 2000). This disparity could be caused by high rates
of gene duplications and losses (Perriere and Gouy 1996). The
same trend was observed in a previous study of Drosophila
(Makino and Kawata 2012) in which many duplicated gene
pairs tended to have the low number of substitutions per
synonymous site (KS) <0.1. Many nondivergent duplicated
1500
1000
500
0
-500
-0.15 -0.10 -0.05 0.00 0.05 0.10
Contrasts in proportion of lost duplicated genes
Mammalian Duplicated Genes and Habitat Variability . doi:10.1093/molbev/msu128
gene pairs would be found as a result of recent duplication
events or assembly errors in a genome. Therefore, we
collapsed recent duplicated genes based on their sequence
similarity. We estimated KS between duplicated genes and
their closest paralogs in mammals and found an apparent
distribution bias toward recent duplication (supplementary
fig. S2, Supplementary Material online). We accordingly col-
lapsed duplicated gene pairs with a KS <0.1 into single genes
(see Materials and Methods). Although there was no corre-
lation between the proportion of recently duplicated genes
and the climatic envelope (supplementary fig. S3A,
Supplementary Material online), there was a significant cor-
relation between PD and the climatic envelope after collapsing
recently duplicated genes (R2 = 0.31, P = 0.013; supplementary
fig. S3B, Supplementary Material online). These results indi-
cate that recently duplicated genes do not explain the habitat
variability of species because their duplication is too recent to
have generated genetic variation, otherwise assembly errors
might disturb the relationship. Note that even when we set
some threshold values of KS (0.010.10) for collapsing dupli-
cated genes, our result did not change (data not shown).
Some mammalian species habitats have been expanded
by human activity. Among species used in this study, Mus
musculus in particular has spread worldwide. When the same
analysis was conducted after the removal of M. musculus, the
trends did not change (the correlation between PD and the
climatic envelope: R2 = 0.35, P = 0.011; the correlation be-
tween PD and the climatic envelope after collapsing recently
duplicated genes: R2 = 0.35, P = 0.012; supplementary fig. S4,
Supplementary Material online). We did not use M. musculus
in the subsequent analyses, although our conclusions were
unaffected by the inclusion of this species.
Some other factors might have affected our results.
Although a correlation between effective population size
and genomic content has been reported (Petit and
Barbadilla 2009), the difference in PD among species was
not related to effective population sizes in Drosophila species
(Makino and Kawata 2012). For the mammalian species used
in this study, we could not estimate effective population size;
this correlation should be evaluated in future studies.
Differences in genome sequence coverage among species
might have influenced our results; however, there was no
correlation between genome sequence coverage and PD (sup-
plementary fig. S5, Supplementary Material online) as shown
in a previous study (Makino and Kawata 2012).
Table 1. Lineage-Specific Evolutionary Events between Closely Related Species.
Species Outgroup Habitat Variability
Duplicated Genes (PD)
Climatic Habitat
Envelope Diversity
Macaca mulatta Callithrix 590 2.233
Pongo abelii jacchus 45 0 0.77
P value
Mus musculus Ochotona 1,201 3.067
Spermophilus princeps 155 1.718
tridecemlineatus
P value
MBE
To reinforce the above results, the habitat diversity defined
by KoppenGeiger climate classification was used instead of
the climatic envelope. This classification is based not only on
temperature and precipitation but also on vegetation (Huijser
1999). Environmental diversity within habitat was estimated
using the Brillouin index (Pielou 1975; Kottek et al. 2006).
When this habitat diversity was used as a measure of envi-
ronment variability, a similar result was obtained (R2 = 0.23,
P = 0.034; supplementary fig. S6A, Supplementary Material
online). The result did not change when PD after collapsing
recently duplicated genes was used (R2 = 0.22, P = 0.037, sup-
plementary fig. S6B, Supplementary Material online). This
result indicates that PD is correlated with habitat variability
not only in Drosophila species but also in euarchontoglires.
Evolutionary Process on Divergence of Duplicated
Genes
We have already shown that recently duplicated genes were
not enriched in species with high habitat variability (supple-
mentary fig. S3A, Supplementary Material online). To identify
evolutionary processes responsible for differences in PD
between species with high and low habitat variability, we
investigated whether in euarchontoglire species a loss of
duplicated genes occurred in species with low PD more
frequently than in those with high PD. Among these species,
there was a significant negative correlation between the
proportion of lost duplicated genes and the climatic envelope
(R2 = 0.51, P = 0.0030; fig. 1B). The result is consistent with a
previous report on Drosophila (Makino and Kawata 2012) in
which species with low habitat variability tended to lose
duplicated genes during evolution.
Overrepresented gene functional categories were
examined for lost genes by Gene Ontology (GO) analysis
(see Materials and Methods) with two closely related species
pairs (Pongo abeliiMacaca mulatta and Spermophilus tride-
cemlineatusM. musculus) in euarchontoglires. Pongo abelii
and S. tridecemlineatus are species with low PD and habitat
variability, whereas Ma. mulatta and M. musculus are species
with high PD and habitat variability (table 1). The results
showed little enrichment of functional categories for genes
in species with low PD; the lost duplicated genes in P. abelii
were enriched in catabolic process (P = 3.6  10 5). There
were no enriched gene functions in S. tridecemlineatus. This
finding indicates that both loss and relaxation of functional
Proportion of Number of Lineage- Proportion of Lineage-
Specific Lost Genes Specific Duplicated Genes
0.81 1,312 917/1,312
1,412 1,128/1,412
2.23  10 9
0.81 584 212/584
0.68 401 282/401
16
<2.2  10
1781
Tamate et al. . doi:10.1093/molbev/msu128
constraints are common for genes in species with low habitat
variability, rather than being gene-specific. For adapting to
heterogeneous environments, species might require not
only some particular biological functions such as cold and
desiccation tolerance but also physiological, morphological,
behavioral, and other adaptive functions. The present results
support the earlier conclusion (Makino and Kawata 2012)
that duplicated genes have been lost from the genome in
species with low habitat variability, whereas high PD has
been maintained in the genome in species with high habitat
variability.
Influence of Different Duplication Mechanisms
When the effects of ohnologs and SSD genes on habitat
variability were separately examined, the proportion of SSD
genes significantly correlated with habitat variability
(R2 = 0.39, P = 0.013; fig. 2A), whereas the proportion of ohno-
logs did not (fig. 2B). The results suggest that the proportion
of SSD genes makes a larger contribution to adaptation to
variable environments than the proportion of ohnologs
although PD calculated using whole genes could be a factor
that restricts adaptation to environmental variation as shown
in Drosophila species (Makino and Kawata 2012). Ohnologs
would be unlikely to be lost under dosage sensitivity
regardless of species habitat variability (Makino and
McLysaght 2010). In contrast, species with high habitat vari-
ability could maintain SSD genes more efficiently, whereas
species with low habitat variability would lose SSD genes
more frequently. As a result, species that had lost SSD
genes might be unable to generate enough genetic variation
for adaptation to heterogeneous environments, leading to
failure to expand their habitat range, resulting in further
losses of SSD genes. Thus, SSD genes can be both a cause
and effect of a change in habitat variability. SSD genes are
MBE
related to response to biotic stimuli for adaptive traits and
were overrepresented in endoderm tissues directly exposed
to external environments (Blomme et al. 2006; Satake et al.
2012). Thus, retained SSD genes that have responded to
various environmental stimuli have played more significant
roles in expanding habitat ranges and in adaptation to novel
environments than ohnologs.
Pairwise Comparison for Mammals
In this study, no correlation between PD and habitat variability
was observed when 30 mammalian species were used. But a
positive correlation might be observed when species within
the same taxonomic group (e.g., species in the same order)
were compared, as in Drosophila (Makino and Kawata 2012)
and euarchontoglires. To investigate the relationship between
PD and habitat variability for other mammalian taxa, pairs of
species were classified by shared membership in an order, a
related order, or a superorder (supplementary table S2,
Supplementary Material online). Figure 3A depicts the rela-
tionship between PD and the climatic envelope in each spe-
cies pair. We found that within pairs, most species with high
PD had higher climatic envelopes than species with low PD
(exact binomial test, P = 0.011). In other words, most closely
related species pairs of mammalian species showed a positive
correlation between PD and habitat variability. Figure 3B de-
picts a similar analysis of SSD genes and the climatic envelope,
showing a significant correlation (exact binomial test,
P = 0.013; fig. 3B). In contrast, there was no significant corre-
lation between the proportion of ohnologs and the climatic
envelope (exact binomial test, P = 0.18; fig. 3C). These results
suggest that PD, particularly of SSD genes, could be a general
index of species capacity to adapt to habitat variability. Rattus
norvegicus has high habitat variability (fig. 3), and it might
affect to the above results. Although the same analyses were

A B
1500
1500

1000
1000
Climatic envelope

Climatic envelope
500
500

0
0

-500
-500

-0.2 -0.1 0.0 0.1 0.2 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20

Contrasts in proportion of SSD gene Contrasts in proportion of ohnolog

FIG. 2. Relationship between the climatic envelope and proportions of different modes of duplicated genes for euarchontoglires including Primates,
Rodentia, and Lagomorpha. (A) Relationship between the climatic envelope and the proportion of SSD genes. The x-axis represents PICs in the
proportion of SSD genes for 16 euarchontoglire species. The y-axis represents PICs in the climatic envelope estimated from WORLDCLIM data sets.
(B) Relationship between the proportion of ohnologs and the climatic envelope. The x-axis represents PICs in the proportion of ohnologs for 16
euarchontoglire species. The y-axis represents PICs in the climatic envelope estimated from WORLDCLIM data sets.

1782
Mammalian Duplicated Genes and Habitat Variability . doi:10.1093/molbev/msu128 MBE

1000
A B

1000
rat rat

800
800
Climatic envelope

Climatic envelope
600
nov
600
nov

luc luc

400
400

por por
cun
pri cun pri
hof ord hof
ord
200

afr afr

200
tri vam cap
eur cap eur vam
ara tri ara hap
hap
str str
tel tel

0
0

0.70 0.75 0.80 0.35 0.40 0.45 0.50

Proportion of duplicated genes Proportion of SSD duplicated genes
1000

C rat
800
Climatic envelope
600

nov

luc
400

cun por

hof ord
afr
pri
cap
200

ara eur tri vam

hap

tel str
0

0.30 0.32 0.34 0.36 0.38 0.40

Proportion of ohnologs
FIG. 3. Relationship between PD and the climatic envelope among closely related mammalian species. Each species is labeled in each plot, and open
squares, closed squares, open circles, closed circles, open triangles, closed triangles, and diamond shapes represent Choloepus hoffmanni (hof)Dasypus
novemcinctus (nov), Myotis lucifugus (luc)Pteropus vampyrus (vam), Erinaceus europaeus (eur)Sorex araneus (ara), Cavia porcellus (por)Dipodomys
ordii (ord)Rattus norvegicus (rat)Spermophilus tridecemlineatus (tri), Oryctolagus cuniculus (cun)Ochotona princeps (pri), Echinops telfairi (tel)
Loxodonta Africana (afr)Procavia capensis (cap), and Haplorrhini (hap)Strepsirrhini (str), respectively. (A) Relationship between PD and the climatic
envelope for pairs of mammalian species sharing an order or suborder. The x-axis represents PICs in the PD for 26 mammalian species. The y-axis
represents PICs in the climatic envelope estimated from WORLDCLIM data sets. A solid line indicates a positive relationship between PD and the
climatic envelope and a dashed line indicates a negative relationship. (B) Relationship between the proportion of SSD genes and the climatic envelope
for pairs of mammalian species sharing an order or suborder. The x-axis represents PICs in the proportion of SSD genes for 26 mammalian species. The y-
axis represents PICs in the climatic envelope estimated from WORLDCLIM data sets. A solid line indicates a positive relationship between the
proportion of SSD genes and the climatic envelope and a dashed line indicates a negative relationship. (C) Relationship between the proportion of
ohnologs and the climatic envelope for pairs of mammalian species sharing an order or suborder. The x-axis represents PICs in the proportion of
ohnologs for 26 mammalian species. The y-axis represents PICs in the climatic envelope estimated from WORLDCLIM data sets. A solid line indicates a
positive relationship between the proportion of ohnologs and the climatic envelope and a dashed line indicates a negative relationship.

conducted after the removal of R. norvegicus, the trends did Prospective Index for Evaluating Evolvability
not change (exact binomial test, P = 0.039 in fig. 3A, P = 0.048 The present results using fully sequenced genomes of mam-
in fig. 3B and P = 0.55 in fig. 3C). Exceptionally, there were malian species support our hypothesis that PD is correlated
some species that showed negative relationships between PD with environmental variability within the species habitat, as
and the climatic envelope (Myotis lucifugusPteropus vam- previously shown for Drosophila. Furthermore, our results
pyrus and Echinops telfairiProcavia capensis; fig. 3A). showed that it is mainly the proportion of SSD genes
Estimation of environmental variability would reflect only (rather than that of ohnologs) that contributes to this corre-
broad-scale patterns. In addition, the measure of habitat lation. Our results also showed that species with low habitat
ranges of species is not perfect. variability and low PD tended to lose duplicated genes.

1783
Tamate et al. . doi:10.1093/molbev/msu128
We suggest that the proportion of SSD genes may be both a
cause and an effect of habitat variability in mammalian
species although further studies would be needed to explore
how maintained SSD genes generate a capacity for adaptation
to various environments. Whole genomes can now be
sequenced relatively easily and the sequences have
been accumulated in databases, so that in the near future
we will be able to estimate PD for thousands of species
(Genome 10 K, https://genome10k.soe.ucsc.edu, last accessed
April 14, 2014 and i5k Insect http://www.arthropodgenomes.
org/wiki/i5K, last accessed April 14, 2014). PD, in particular the
proportion of SSD genes, would be an appropriate index for
evaluating species vulnerability to future environmental
changes for the conservation of biodiversity.
Materials and Methods
Sequences of Fully Sequenced Mammalian Species
Nucleotide and protein sequences corresponding to protein-
encoding genes from 30 mammalian species (Primates:
Callithrix jacchus, Gorilla gorilla, Ma. mulatta, Microcebus mur-
inus, Nomascus leucogenys, Otolemur garnettii, Pan troglodytes,
P. abelii, and Tarsius syrichta; Rodentia: Cavia porcellus,
Dipodomys ordii, M. musculus, R. norvegicus, and S. tridecem-
lineatus; Lagomorpha: Ochotona princeps and Oryctolagus
cuniculus; Others: Ailuropoda melanoleuca, Choloepus
hoffmanni, Dasypus novemcinctus, E. telfairi, Erinaceus
europaeus, Loxodonta africana, Macropus eugenii,
Monodelphis domestica, My. lucifugus, Ornithorhynchus
anatinus, Pr. capensis, Pt. vampyrus, Sorex araneus, and
Vicugna pacos) were downloaded from the Ensembl database,
release 63 (Flicek et al. 2011).
Duplicated Genes
To identify duplicated genes in mammalian genomes, we
conducted an all-against-all Basic Local Alignment Search
Tool (BLAST) search for all protein sequences, and we used
the longest sequence for genes with multiple isoforms. Genes
with a homolog (E value <10 5 and query coverage >30%) in
the same species were identified as candidate duplicated
genes. In particular, with respect to E value, it was found in
a previous study that the use of different E value cutoffs (10 5,
10 10, and 10 20) did not change the results (Makino and
Kawata 2012). PD was defined as the ratio of the number of
duplicated genes to the total number of genes.
Synonymous Substitution Rate between Duplicated
Genes
To investigate the distribution of gene duplication timing for
the duplicated genes, we aligned the sequences of the dupli-
cated gene pairs derived from the Ensembl database using the
ClustalW2 multiple sequence alignment program (Larkin
et al. 2007) and estimated the synonymous substitution
rate (Ks) between a duplicated gene and its closest paralog
by the Yang and Nielsen method (2000), implemented in the
Phylogenetic Analysis by Maximum Likelihood program pack-
age (Yang 1997). A closest paralog was identified as the best
1784
MBE
BLAST hit among duplicated gene candidates (supplemen-
tary tables S1 and S2, Supplementary Material online).
Gene Collapse Based on Sequence Similarity
Homologous gene pairs with Ks <0.1 were collapsed into
single genes (a2a3a4 and b1b2 in supplementary fig.
S7A, Supplementary Material online). The collapsed genes
were classified as duplicated genes on the basis of a BLAST
hit in comparison with genes not recently duplicated. If at
least one gene in the homologous gene cluster (a2a3a4 in
supplementary fig. S7A, Supplementary Material online) had
a duplicated gene partner (a1 in supplementary fig. S7A,
Supplementary Material online), the collapsed gene was de-
fined as a duplicated gene. If not (b1b2 in supplementary fig.
S7A, Supplementary Material online), the collapsed gene was
defined as a singleton.
Ohnologs and SSD Genes
We used 7,294 human ohnologs from Makino and McLysaght
(2010). To obtain mammalian ohnologs (derived from WGD)
and SSD duplicated genes, we downloaded orthologs of
human genes for each mammalian species from Ensembl
database release 63 (Flicek et al. 2011). If an ortholog corre-
sponded to a human ohnolog (Makino and McLysaght 2010),
the gene was defined as a candidate ohnolog (supplementary
table S1, Supplementary Material online). When the candi-
date ohnolog were species-specific singletons, they were ex-
cluded. Duplicate genes not including ohnologs were defined
as SSD genes.
Lineage-Specific Gene Losses
Orthologs were defined by reciprocal best hits between dif-
ferent species using the results of the all-against-all BLAST. If
the orthology relationship was obtained from one-to-one
best hits, the ortholog was defined as one-to-one. Such a
relationship indicated that there had been no lineage-specific
gene duplication and loss after speciation. Orthologous gene
clusters were identified using one-to-one orthologous rela-
tionship for closely related species and their outgroups to
investigate gene loss events during evolution (supplementary
fig. S7B, Supplementary Material online). We did not use
genes without orthologs in the outgroups because their an-
cestral state could not be determined. If there was no gene
loss event in either of the closely related species, we obtained
orthologous trios where possible (e.g., species1Bspecies2B
outgroupB in supplementary fig. S7B, Supplementary
Material online). When orthologous trios were not available,
gene loss events occurring in either lineage of the closely
related species were inferred (supplementary fig. S7B,
Supplementary Material online). To identify a species lost
duplicated genes generated before speciation from another
species, we focused on the gene similarity between a species
and the outgroup species (supplementary fig. S7B,
Supplementary Material online). A species (e.g., species1 in
supplementary fig. S7B, Supplementary Material online) was
defined as having a lost duplicated gene (specie1B in supple-
mentary fig. S7B, Supplementary Material online) when the
Mammalian Duplicated Genes and Habitat Variability . doi:10.1093/molbev/msu128
following were observed: an inferred ortholog (species2A in
supplementary fig. S7B, Supplementary Material online) in
the compared species (species2 in supplementary fig. S7B,
Supplementary Material online) was a duplicated gene and
the similarity between the duplicated gene and its duplicated
gene partner (similarity between species2A and species2B in
supplementary fig. S7B, Supplementary Material online) was
lower than that between either of the duplicated copies and
its best-hit homolog in the outgroup (similarity between
species2A and outgroupA or between 2B and outgroupB in
supplementary fig. S7B, Supplementary Material online), as
determined by BLAST search.
Habitat Area and Variability
The habitat areas for mammalian species were obtained from
the International Union for Conservation of Natural
Resources (IUCN) Red List of Threatened Species and litera-
tures (http://www.iucnredlist.org, last accessed April 14, 2014)
(Huijser 1999; Marroig et al. 2004). Habitat variability was
estimated from the climatic envelope, and habitat diversity
was estimated using the Koppen climatic classification
(Kottek et al. 2006). The climatic envelope is the range of
temperatures, precipitation, and other climate-related param-
eters in which a species is currently found (supplementary
table S3, Supplementary Material online). We estimated the
climatic envelope using principal component analysis (PCA)
with WORLDCLIM (Hijmans et al. 2005). We obtained world
special data and the WORLDCLIM climatic data set (10 min
latitude/longitude) from DIVA-GIS (http://www.diva-gis.org,
last accessed April 14, 2014). The habitat area was measured
as the number of grid squares on the climate map. We then
extracted the climatic values from 19 bioclimatic variables
used for BIOCLIM (Hijmans et al. 2005) in the habitat area
of each mammalian species. We performed PCA using the
bioclimatic variables for all of the species and found that the
first two principal components (PCs) explained 93.6% of the
total variance (supplementary table S3, Supplementary
Material online). The contributions of PC1 and PC2 were
77.9% and 15.7%, respectively. PCA plots (x-axis: PC1; y-axis:
PC2) are shown in supplementary figure S8, Supplementary
Material online. On the basis of the PCA results, PC1 and PC2
values were plotted for each species. A total of 122,303 cells
were used (PC1: 779  PC2: 157) by weighting the relative
contributions to PC1 and PC2 for estimating the climatic
envelope, and the numbers of cell grid overlapping points
in the 122,303 cell grids were defined as the climatic envelope
of mammalian species.
The Koppen climatic classification map was used for esti-
mating mammalian species habitat diversity (Kottek et al.
2006). This climate map consists of a grid of squares (0.5
latitude/longitude) in which a certain climate is classified by
temperature, precipitation, and vegetation. The habitat area
was measured as the number of grid squares on the climate
map. The number of grids varied among the mammalian
species, and accordingly, habitat diversity was calculated
using varieties (through logarithmic transformation) in the
climatic environment among grid squares for each species
MBE
using Brillouins index, which is robust to sample size
(Margalef 1958).
Model Selection
All of the following statistical analyses were executed with R
software (2.13.2 version) (http://www.r-project.org, last
accessed April 14, 2014). Model selection was applied using
linear models to determine which genomic factors affect hab-
itat features. The set of predictors of the explanatory variables
that yielded the lowest Akaike Information Criterion (AIC)
was selected using a stepwise AIC procedure. PD, PD for SSD
genes, PD for ohnologs, genome size, the number of genes, and
the number of duplicated genes as genomic factors were
used, and the climatic envelope, habitat diversity, diet
breadth, and habitat area were used as habitat features.
Diet breadth is the number of dietary categories eaten by
each species measured using any qualitative or quantitative
dietary measure over any period of time using any assessment
method for noncaptive or nonprovisioned populations (Jones
et al. 2009).
To remove any phylogenetic constraints on the relation-
ship between genetic architecture and habitat, phylogenetic
trees and phylogenetic independent contrasts (PICs) were
inferred. Phylogenetic trees were inferred using matrix extra-
cellular phosphoglycoprotein precursor gene (Bardet et al.
2010). The ClustalW2 multiple sequence alignment program
(Larkin et al. 2007) was used for alignment and NJplot was
used for constructing phylogenetic trees (Perriere and Gouy
1996). Using the phylogenetic trees, we selected a model by
applying a generalized least squares model with a Brownian
model, as described earlier, and measured the PICs
(Felsenstein 1985). The model selection was applied using
the estimated PICs.
To evaluate the relationship between PD and habitat var-
iability in species paired by common order, common super-
order, and related order, we performed pairwise comparisons
using a sign test.
Gene Ontology
To determine whether lineage-specific lost duplicated
genes of species with low PD were enriched in specific
functional categories, we examined the GO database
entries for the duplicated genes between species with differ-
ent PD (C. jacchusP. abelii and M. musculusS. tridecemlinea-
tus) (table 1). The GO identifiers (IDs) and GO slim
biological process annotations for M. musculus and Homo
sapiens were downloaded from ftp://ftp.geneontology.org/
pub/go/gene-associations/ (last accessed April 14, 2014) and
http://www.geneontology.org/GO_slims (last accessed April
14, 2014), respectively. The frequency of each GO ID was
counted for genes from M. musculus and H. sapiens. For the
other species, we used the GO IDs of the most similar homo-
log in M. musculus (for Rodentia and Lagomorpha) and H.
sapiens (for Primates). The enrichment of GO IDs for genes in
species with low PD was compared with that in species with
high PD. We calculated the P value for each GO ID by
1785
Tamate et al. . doi:10.1093/molbev/msu128
comparing the two different gene sets. The estimated P values
were adjusted by Bonferroni correction.
Supplementary Material
Supplementary tables S1S3 and figures S1S8 are available at
Molecular Biology and Evolution online (http://www.mbe.
oxfordjournals.org/).
Acknowledgments
The authors thank our laboratory members for their helpful
advice. This work was supported by KAKENHI (25291096)
from the Japan Society for the Promotion of Science to
T.M., and partly by the Global COE Program Centre for
ecosystem management adapting to global change (J03) of
the Ministry of Education, Culture, Sports, Science and
Technology of Japan (MEXT) to M.K.
References
Bardet C, Delgado S, Sire JY. 2010. MEPE evolution in mammals reveals
regions and residues of prime functional importance. Cell Mol Life
Sci. 67:305320.
Birchler JA, Riddle NC, Auger DL, Veitia RA. 2005. Dosage balance in
gene regulation: biological implications. Trends Genet. 21:219226.
Birchler JA, Veitia RA. 2012. Gene balance hypothesis: connecting issues
of dosage sensitivity across biological disciplines. Proc Natl Acad Sci
U S A. 109:1474614753.
Blomme T, Vandepoele K, De Bodt S, Simillion C, Maere S, Van de Peer
Y. 2006. The gain and loss of genes during 600 million years of
vertebrate evolution. Genome Biol. 7:R43.
Brunet FG, Crollius HR, Paris M, Aury JM, Gibert P, Jaillon O, Laudet V,
Robinson-Rechavi M. 2006. Gene loss and evolutionary rates follow-
ing whole-genome duplication in teleost fishes. Mol Biol Evol. 23:
18081816.
Dean EJ, Davis JC, Davis RW, Petrov DA. 2008. Pervasive and persistent
redundancy among duplicated genes in yeast. PLoS Genet. 4:
e1000113.
Dehal P, Boore JL. 2005. Two rounds of whole genome duplication in the
ancestral vertebrate. PLoS Biol. 3:e314.
Felsenstein J. 1985. Phylogenies and the comparative method. Am Nat.
125:115.
Flicek P, Amode MR, Barrell D, Beal K, Brent S, Chen Y, Clapham P,
Coates G, Fairley S, Fitzgerald S, et al. 2011. Ensembl 2011. Nucleic
Acids Res. 39:D800D806.
Hakes L, Pinney JW, Lovell SC, Oliver SG, Robertson DL. 2007. All dupli-
cates are not equal: the difference between small-scale and genome
duplication. Genome Biol. 8:R209.
Hartman JLT, Garvik B, Hartwell L. 2001. Principles for the buffering of
genetic variation. Science 291:10011004.
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high
resolution interpolated climate surfaces for global land areas. Int J
Climatol. 25:19651978.
Hoffmann AA, Blows MW. 1994. Species borders: ecological and evolu-
tionary perspectives. Trends Ecol Evol. 9:223227.
Hufton AL, Groth D, Vingron M, Lehrach H, Poustka AJ, Panopoulou G.
2008. Early vertebrate whole genome duplications were predated by
a period of intense genome rearrangement. Genome Res. 18:
15821591.
Huijser M. 1999. Human impact on populations of hedgehogs Erinaceus
europaeus through traffic and changes in the landscape: a review.
Lutra 41:4954.
Jones KE, Bielby J, Cardillo M, Fritz SA, ODell J, Orme CDL, Safi K,
Sechrest W, Boakes EH, Carbone C, et al. 2009. PanTHERIA: a spe-
cies-level database of life history, ecology, and geography of extant
and recently extinct mammals. Ecology 90:26482648.
1786
MBE
Kellermann V, van Heerwaarden B, Sgro CM, Hoffmann AA. 2009.
Fundamental evolutionary limits in ecological traits
drive Drosophila species distributions. Science 325:12441246.
Kirkpatrick M, Barton NH. 1997. Evolution of a species range. Am Nat.
150:123.
Kottek M, Grieser J, Beck C, Rudolf B, Rubel F. 2006. World map of the
KoppenGeiger climate classification updated. Meteorol Z. 15:
259263.
Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA,
McWilliam H, Valentin F, Wallace IM, Wilm A, Lopez R, et al.
2007. Clustal W and Clustal X version 2.0. Bioinformatics 23:
29472948.
Lynch M, Conery JS. 2000. The evolutionary fate and consequences of
duplicate genes. Science 290:11511155.
Maere S, De Bodt S, Raes J, Casneuf T, Van Montagu M, Kuiper M, Van
de Peer Y. 2005. Modeling gene and genome duplications in eukary-
otes. Proc Natl Acad Sci U S A. 102:54545459.
Makino T, Hokamp K, McLysaght A. 2009. The complex relationship of
gene duplication and essentiality. Trends Genet. 25:152155.
Makino T, Kawata M. 2012. Habitat variability correlates with
duplicate content of Drosophila genomes. Mol Biol Evol. 29:
31693179.
Makino T, McLysaght A. 2010. Ohnologs in the human genome are
dosage balanced and frequently associated with disease. Proc Natl
Acad Sci U S A. 107:92709274.
Makino T, McLysaght A, Kawata M. 2013. Genome-wide deserts
for copy number variation in vertebrates. Nat Commun. 4:2283.
Margalef D. 1958. Information theory in ecology. Gen Syst. 3:36071.
Marroig G, Cropp S, Cheverud JM. 2004. Systematics and evolution of
the Jacchus group of marmosets (Platyrrhini). Am J Phys Anthropol.
123:1122.
McLysaght A, Hokamp K, Wolfe KH. 2002. Extensive
genomic duplication during early chordate evolution. Nat Genet.
31:200204.
Nakatani Y, Takeda H, Kohara Y, Morishita S. 2007. Reconstruction of
the vertebrate ancestral genome reveals dynamic genome reorgani-
zation in early vertebrates. Genome Res. 17:12541265.
Ohno S. 1970. Evolution by gene duplication. New York: Springer.
Perriere G, Gouy M. 1996. WWW-query: an on-line retrieval system for
biological sequence banks. Biochimie 78:364369.
Petit N, Barbadilla A. 2009. Selection efficiency and effective population
size in Drosophila species. J Evol Biol. 22:515526.
Pielou EC. 1975. Ecological diversity. New York: Wiley.
Satake M, Kawata M, McLysaght A, Makino T. 2012. Evolution of ver-
tebrate tissues driven by differential modes of gene duplication.
DNA Res. 19:305316.
Tsuda ME, Kawata M. 2010. Evolution of gene regulatory networks by
fluctuating selection and intrinsic constraints. PLoS Comput Biol. 6:
e1000873.
Veitia RA. 2002. Exploring the etiology of haploinsufficiency. Bioessays
24:175184.
Veitia RA. 2004. Gene dosage balance in cellular pathways: implications
for dominance and gene duplicability. Genetics 168:569574.
Veitia RA. 2003. Nonlinear effects in macromolecular assembly and
dosage sensitivity. J Theor Biol. 220:1925.
Wapinski I, Pfeffer A, Friedman N, Regev A. 2007. Natural history and
evolutionary principles of gene duplication in fungi. Nature 449:
5461.
Wilkins AS. 1997. Canalization: a molecular genetic perspective.
Bioessays 19:257262.
Wolfe K. 2000. Robustnessits not where you think it is. Nat Genet. 25:
34.
Yang Z. 1997. PAML: a program package for phylogenetic analysis by
maximum likelihood. Comput Appl Biosci. 13:555556.
Yang Z, Nielsen R. 2000. Estimating synonymous and nonsynonymous
substitution rates under realistic evolutionary models. Mol Biol Evol.
17:3243.