Download Multiple speciation events in an arthropod with divergent evolution

Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
Proc. R. Soc. B (2010) 277, 689–696
doi:10.1098/rspb.2009.1822
Published online 4 November 2009
Multiple speciation events in an arthropod
with divergent evolution in
sexual morphology
Teiji Sota1,* and Tsutomu Tanabe2
1
Department of Zoology, Graduate School of Science, Kyoto University, Sakyo, Kyoto 606-8502, Japan
2
Faculty of Education, Kumamoto University, Kurokami, Kumamoto 860-8555, Japan
Sexual selection can facilitate divergent evolution of traits related to mating and consequently promote
speciation. Theoretically, independent operation of sexual selection in different populations can lead to
divergence of sexual traits among populations and result in allopatric speciation. Here, we show that
divergent evolution in sexual morphology affecting mating compatibility (body size and genital morphologies) and speciation have occurred in a lineage of millipedes, the Parafontaria tonominea species
complex. In this millipede group, male and female body and genital sizes exhibit marked, correlated divergence among populations, and the diverged morphologies result in mechanical reproductive isolation
between sympatric species. The morphological divergence occurred among populations independently
and without any correlation with climatic variables, although matching between sexes has been
maintained, suggesting that morphological divergence was not a by-product of climatic adaptation.
The diverged populations underwent restricted dispersal and secondary contact without hybridization. The
extent of morphological difference between sympatric species is variable, as is diversity among allopatric
populations; consequently, the species complex appears to contain many species. This millipede case
suggests that sexual selection does contribute to species richness via morphological diversification
when a lineage of organisms consists of highly divided populations owing to limited dispersal.
Keywords: body size; genitalia; millipede; reproductive isolation; sexual selection
1. INTRODUCTION
Since Darwin’s (1871) initial insight into sexual selection,
the idea that sexual selection promotes speciation has
been repeatedly considered (West-Eberhard 1983;
Panhuis et al. 2001; Ritchie 2007). However, the role of
sexual selection in speciation in nature is often obscured
by prominent ecological diversification, with which
some differential mate choice may be associated (Ritchie
2007). Thus, how often sexual selection acts to promote
speciation without the collaboration of natural selection
remains unclear (Hendry 2009). Evolutionary models of
sexual selection predict that male and female sexual
traits related to mating will show correlated evolution
along a line of equilibrium (Lande 1982; Iwasa &
Pomiankowski 1995; Gavrilets 2000). It follows that
speciation owing to divergent evolution of sexual traits
is most likely to occur among allopatric populations
experiencing varying intensities and directions of sexual
selection. However, the relationship between allopatric
divergence in sexual traits by sexual selection and
speciation has not been clearly revealed.
Sexual selection can act on various traits involved in
pre-zygotic stages of sexual reproduction (Andersson
1994). Empirical studies have focused on the divergent
evolution of male courtship signals and female preference,
which constitute major pre-zygotic isolation mechanisms.
However, divergent evolution in morphology, especially
the genitalia of internally fertilized animals, also can be
caused by sexual selection (Eberhard 1985), and the
resultant morphological difference can lead to mechanical
reproductive isolation (Coyne & Orr 2004). Comparative
studies suggest that sexual selection promotes divergence
of genital morphologies (Arnqvist 1998; Takami & Sota
2007), but the relationship between divergence in genital
morphology and speciation has not been well understood.
The effectiveness of species-specific genital morphology
in reproductive isolation has been questioned repeatedly
(Eberhard 1985; Shapiro & Porter 1989). Nevertheless,
cases of parapatric or sympatric species pairs exist, in
which pre-mating isolation is incomplete; yet, mechanical
isolation occurs owing to incompatible genital morphologies, resulting in a high cost of interspecific mating
(Sota & Kubota 1998; Tanabe & Sota 2008). These
cases suggest that divergent genital evolution can occur
rapidly enough to precede the evolution of pre-mating
isolation and to be the primary process leading to
speciation.
Here we elucidate the link between divergent evolution
of mechanical agents of reproductive isolation and speciation in a millipede lineage consisting of highly diverged
populations in sympatry as well as allopatry. The tonominea
species complex of the genus Parafontaria is endemic to
Japan (figure 1c) and exhibits marked variations in body
size, genital size and genital morphology among populations, which hinder delimiting species boundaries
based on morphology (Tanabe et al. 2001). In this species
complex, two different forms occur together in some
* Author for correspondence ([email protected]).
Electronic supplementary material is available at http://dx.doi.org/10.
1098/rspb.2009.1822 or via http://rspb.royalsocietypublishing.org.
Received 6 October 2009
Accepted 15 October 2009
689
This journal is q 2009 The Royal Society
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
690 T. Sota & T. Tanabe Multiple speciation events
(b)
34
°N
(a)
64/0.68
84/1.0
99/1.0
68/0.98
93/1.0
Y31 P26
Y22 P24x2
Y157 P26
84/1.0 Y158 P26
Y89 P28x2
Y327 P28
56/0.81 Y21 P24
50/0.50
Y365 P55x2
96/1.0 Y33 P27
Y240 P73
Y29 P23
Y32 P23x6
Y69
P37
P33
76/0.80 Y278 P36x6
100/1.0 Y70 P37
1 mm
Y340 P37
Y342 P37
75/0.82
96/1.0
Y341 P37
Y102 P32
P34
Y369 P57x2
0
200 km
56/1.0
Y280 P57
male
female
Y39 P33x2
90/1.0
Y51 P33
E
Y52 P33
°
97/1.0
5
Y281 P58x3
13
61/1.0
Y42 P34x2
–/0.99
Y41 P34x2
Y91 P29
Y285 P60x2
Y284 P59
–/0.97
Y81 P38x6
99/1.0
Y84 P39
Y122 P40
100/1.0
Y88 P18
Y121 P19
86/0.92 Y347 P41
Y349 P41
94/1.0
Y76 P41x3
–/0.64
98/1.0
Y75 P41
Y58 P42x3
P35x3
52/– Y330
Y331 P35
100/1.0
Y74 P35
–/0.97
Y73 P35
99/1.0
Y384 P69
E
Y385
P69
0°
P70
98/1.0 Y210
13
Y213 P70
96/1.0
Y212 P70
Y211 P70
94/1.0
Y153 P9
74/0.95
Y155 P9
Y156 P9
90/1.0 Y123 P8
Y124 P10x3
–/0.56
Y154 P9
62/0.86 Y28 P20
Y245 P66
Y244
P66x2
80/0.99 64/0.93
–/0.93
Y248 P66
–/.0.59Y250 P66
99/1.0 Y5 P76
64/.98
Y13 P76
100/1.0 Y120 P76
P42
1 mm
80/1.0
Y86 P17
Y87
P17
–/0.92
Y85 P17
92/1.0
P43
69/0.89Y1 P9
–/0.90
Y214 P71
99/1.0
Y104 P16
Y277 P16
female
male
100/1.0
Y67 P15x2
Y68 P15
98/1.0
65/0.88 Y132 P21x2
(c)
Y135 P21
Y131 P21
Y134 P21
Y27
–/0.77
58/– Y163P21
P22
98/1.0
Y130 P21
Y30 P22
97/1.0
Y56 P49x4
93/1.0
Y77 P49
93/1.0
Y288 P63
Y55 P51
Y60 P43
100/1.0
Y275 P56x3
Y368 P56x2
Y66 P45x2
Y358 P45
100/1.0
Y49 P45
71/0.94
Y65 P45
94/1.0
Y287 P62
Y376 P61x4
73/0.99
94/1.0 Y379 P61
Y286 P61
88/1.0
92/1.0 Y37 P47x2
97/1.0
82/0.99 Y47 P44 Y63 P47
–/0.78
P44
87/0.98
Y48 P44
–/0.72
1 mm
Y64 P44x2
–/0.59
Y355 P44
Y26 P54
94/1.0
Y299 P65
Y383 P68
99/1.0
Y57 P52
99/1.0
Y78 P52
Y289 P64
P45
73/0.87 Y105 P77
100/1.0
Y106 P7x3
Y233 P72x2
male
female
Y36 P46x6
93/1.0 Y103 P4
tonominea species complex
62/0.73
Y323 P4
73/0.99
Y101 P14
95/1.0
Y98 P5
53/1.0
99/1.0
Y99 P2
Y100 P3
100/1.0
–/0.75
Y3 P75x2
Y92 P74
Y23 P6
Y10 doenitzi
98/1.0
Y290 doenitzi
laminata
89/1.0 73/0.95 Y119
–/0.79
Y315 laminata
98/1.0
Y229 laminata
85/1.0
Y4 laminata
Y126 laminata
100/1.0
Y43 crenata
94/1.0
69/1.0
Y93 crenata
–/0.72
Y2 longa
–/0.54
Y216 longa
Parafontaria
79/1.0
Y221 longa
0.005 substitutions per site
Y96 longa
Y97 longa
98/1.0
Y9 erythrosoma
98/1.0
Y94 erythrosoma
Y11 ishiii
64/0.99
Y110 R. cornuta
75/1.0
Y111 R. semicircularis
Riukiaria
74/1.0
Y215 R. semicircularis
73/1.0
Y112 R. chelifera
outgroup
Xystodesmus
Y46 X. serrulatus
100/1.0
Y6 L. takakuwai
Levizonus
Y8 L. montanus
14
0°
E
97/1.0
Figure 1. (a) Sampling sites of the tonominea species complex in Japan. Open circles are single-species sites. Coloured circles are
two-species sites. (b) Phylogenetic tree of Parafontaria millipedes resulting from a maximum-likelihood analysis of mitochondrial COI–
COII sequences. Node supports are indicated by bootstrap percentages (when .50%) and Bayesian posterior probabilities. Sympatric
species pairs are connected by coloured lines corresponding to the colours in (a). Differences in genital morphologies between sympatric
species are illustrated for three example sites. (c) Males of sympatric species of the tonominea species complex from Rurikei, Kyoto.
Proc. R. Soc. B (2010)
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
Multiple speciation events
localities, and differences between two sympatric forms
are evidenced by morphological discontinuity with an
absence of intermediate forms indicating hybridization
(Tanabe et al. 2001). Furthermore, experimental evidence
has indicated mechanical isolation between sympatric
forms owing to body size and genital mismatch (Tanabe &
Sota 2008). The mating behaviour in this millipede
group consists of alignment of the genitalia between
sexes, preliminary intromission of the gonopods without
sperm transfer and true intromission with sperm transfer
(Tanabe & Sota 2008). Mismatches in body and genital
size/shape hinder these copulatory behaviours because
body length differences prevent alignment of genital
openings, and genital size/shape differences prevent
proper intromission of male gonopods into female genital
cavities (Tanabe & Sota 2008). Corresponding structures
in male and female devices may function in interactions
between sexes and may be subject to coevolution by
sexual selection. Notably, Parafontaria millipedes are detritivores that consume leaf litter and are confined to forest
floors in the temperate zone. No ecological divergence is
recognized among the members of Parafontaria. Thus, the
tonomiea species complex provides an intriguing case in
which to examine the relationship between sexual selection and speciation, possibly excluding the predominant
effect of ecologically divergent selection.
We examined whether the divergence pattern in the
tonominea species complex adheres to the model of allopatric speciation by sexual selection. We focused on body
and genital sizes of both sexes as key mating traits, and
conducted analyses of genetic and morphological
differentiation of the millipede populations using mitochondrial and nuclear gene sequences. We show that
population divergence in this millipede lineage has
occurred as a result of contiguous dispersion and isolation, as would be expected from the low mobility of
millipedes; morphological divergence has not been
affected by environmental clines and has occurred without interdependency among neighbour populations, as
would be expected from allopatric divergence by sexual
selection; and by assessing phylogenetically independent
contrasts, correlated evolution in mating traits between
sexes has occurred to maintain matching between sexes,
whereas the evolution of genital traits was partly independent from that in body size, providing broader phenotypes
to produce mechanically isolated species.
2. MATERIAL AND METHODS
(a) Sampling
The P. tonominea species complex (including P. falcifera and
P. tokaiensis) was sampled from 54 locations, 14 of which harboured two sympatric species, in Honshu, Shikoku and
Kyushu, Japan (figure 1a; table S1a in the electronic supplementary material). Other Parafontaria and outgroup taxa
from four xystodesmid genera (Riukiaria, Xystodesmus and
Levizonus) were collected in Hokkaido through Ryukyu,
Japan (table S1b in the electronic supplementary material),
for phylogenetic analysis.
(b) Molecular methods
Total genomic DNA was extracted from legs preserved in 99
per cent or 80 per cent ethanol. To amplify the mitochondrial
COI –COII gene region, we used primers COS2183N
Proc. R. Soc. B (2010)
T. Sota & T. Tanabe
691
(50 -CAR CAY YTA TTY TGR TTY TTY GG-30 ) and
COA3662 (50 -CCA CAA ATT TCT GAA CAT TGA CC-30 ),
and additional primers in between COA2745Y (50 -T YAA
MCC YAA RAA ATG YTG AG-30 ); COSY1 (50 -GCT
ATT ACT TTT TGR TTT CC); COAY1 (50 -CGA TTY
AAY ATY ATA TTT CT-30 ); and COAY2 (50 -ACA GCA
TCY AAY TTW ACY CC-30 ). The data matrix consisted
of 1275 bp sequences that are unambiguously aligned.
About 600 bp of the nuclear elongation factor 1-alpha gene
(EF-1a) was amplified using primers DiploEF1aF (50 -GCC
TGG GTT TTG GAT AAA CTT AAG GC-30 ) and DiploEF1aR3 (50 -CCT CCA ATC TTG TAA ACG TC-30 ).
Sequences have been deposited in GenBank (accession numbers: FJ775184–FJ775328; FJ775329 –FJ775526). The
aligned data matrix had 614 sites with a 519 bp exon and
95 bp (including gaps) intron sequence. For COI –COII
and EF-1a data, maximum likelihood (ML) trees were
obtained by a genetic algorithm implemented in GARLI
v. 0.96 (http://www.bio.utexas.edu/faculty/antisense/garli/
Garli.html). The GTR þ I þ G substitution model was
used, and the credibility of nodes was assessed by 1000 bootstrap replicates. Bayesian analyses with MRBAYES v. 3.1.1
(Huelsenbeck & Ronquist 2001) were also conducted to
obtain posterior probabilities that different nodes appeared
in the ML trees. Also, the average sequence difference
between populations was obtained from ARLEQUIN v. 3.11
(Excoffier et al. 2005) for use in Mantel tests.
(c) Estimation of divergence time
Divergence times of populations within the tonominea
species complex were estimated using mitochondrial gene
sequences from eight species outside the tonominea species
complex and 66 tonominea species complex sequences. An
ML tree was constructed using GARLI, and the divergence
times were estimated using a relaxed clock model using
MULTIDIVTIME (Thorne & Kishino 2002). Since no fossil
or geohistoric evidence was available for calibrating the millipede divergence time, we referred to the evolutionary rate of
the COI gene of Plateumaris beetles in Japan (Sota &
Hayashi 2007), because the sequence region in the
Plateumaris data was identical to the early part of the millipede sequence data and both organisms occur in the same
geographical region. Accounting for a time-dependent rate
of sequence divergence (Ho et al. 2005), the rate of
sequence divergence per million years (Myr) at t Myr BP
was expressed as 0.032 exp(21.984t) þ 0.018. Divergence times corresponding to the sequence divergences
(uncorrected pairwise distances) predicted under this evolutionary rate were used to set priors and constraints for
nodes and the evolutionary rate. The root node of Parafontaria had a mean sequence divergence corresponding to
3.6 Myr, which was used for rttm, the mean of the prior distribution for the age of the ingroup root (with s.d., rttmsd of
1.8 Myr). Both the mean and s.d. (rtrate, rtratesd) of the
prior distribution of the evolutionary rate at the root node
were set at 0.016. The sequence divergence at the root
node of the tonominea species complex corresponded to
2.7–2.1 Myr and this range was used to constrain the root
node. Regarding 11 nodes for the coalescence of sympatric
species, we set conservative lower and upper ages as the
mean age + 0.5 mean age, where the mean age was estimated from the mean sequence divergence between
sympatric species using the above equation.
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
692 T. Sota & T. Tanabe Multiple speciation events
(a) 0.12
(e) Statistical methods
The relationship between geographical distance and genetic
distance between populations was tested with a Mantel test
implemented in the R-package (Casgrain & Legendre
2004). An extension of the Mantel test, the partial matrix
correspondence test (Manly 1986) implemented in RTMANT v. 2.1 (Manly 1997) was used to determine the contributions of geographical distance, genetic distance and
morphological difference to variation in genital morphology.
A Mantel test for correlations among mitochondrial genetic
distance, geographical distance and morphological distance
between populations was performed with the R-package.
To assess environmental factors affecting body and genital
size variation, stepwise multiple regression was performed
using three geographic (latitude, longitude and altitude)
and two climatic (annual mean temperature and annual rainfall) variables. These variables potentially show significant
correlations with arthropod body size (e.g. Masaki 1967;
Mousseau 1997; Sota et al. 2000; Bidau & Martı́ 2008).
To study the evolutionary rates and correlated evolution of
sexual body size and genital sizes, phylogenetically independent
contrasts (Felsenstein 1985) were obtained based on the ultrametric mitochondrial tree using the PDAP module (Midford
et al. 2003) in MESQUITE v. 2.6 (Maddison & Maddison
2004). Thirty-six populations with all four morphological
measurements were included in this analysis. All measurements
were log10-transformed. To adjust the dimension of measurements in the comparison of evolutionary rates, genital weight
(but not metatergal width) was divided by 3 after log10 transformation. To standardize independent contrasts adequately,
the branch lengths were log-transformed (Garland et al.
1992). To assess heterogeneity in the evolutionary rates
among characters, a Friedman test was used, and absolute
values of standardized contrasts at different nodes were
compared with a Wilcoxon sign test (Takami & Sota 2007).
0.06
3. RESULTS
(a) Genetic divergence among populations
and divergence time
Mitochondrial and nuclear gene genealogies provided evidence for genetic differentiation among the sympatric and
allopatric populations of the tonominea species complex.
In the COI–COII gene sequence (figure 1b), the tonominea
species complex consists of diverged haplotypes (maximum pairwise sequence divergence, 10.27%), and
sympatric species pairs showed 0.45–7.69% sequence
divergence. The gene tree indicates that sympatric species
were not sisters in most cases. The nuclear EF-1a gene
(exonþintron) sequences showed lower divergence (maximum pairwise sequence divergence 2.36%; figure S1a in
the electronic supplementary material), but sympatric
species had sequences with 0.24–0.77% difference.
Within the tonominea species complex, the genetic distance
between populations in mitochondrial and nuclear genes
increased significantly with geographical distance (figure 2;
Proc. R. Soc. B (2010)
genetic distance between populations
(d) Morphological analysis
Male and female body sizes were expressed as the tenth
metatergal width. Dried genital organs (male gonopods and
female genital segment) were weighed. Genital weight is closely correlated with other elaborate measures of genital
dimensions such as the length of the male genital appendage
and centroid size of female genitalia (Tanabe et al. 2001).
0.10
0.08
0.04
0.02
0
genetic distance between populations
(b) 0.04
0.03
0.02
0.01
0
200
400
600
800
1000
geographic distance between populations (km)
Figure 2. Relationship between geographical and genetic distance in the tonominea species complex. (a) COI –COII gene
sequence. (b) EF-1a gene sequence. Genetic distance is the
uncorrected proportion of sequence difference.
Mantel test; COI–COII, r ¼ 0.446, p ¼ 0.0001; EF-1a,
r ¼ 0.522, p ¼ 0.0001), indicating restricted gene flow and
isolation by distance among geographical populations.
The common ancestor of the tonominea species complex was estimated to have occurred 2.4 Myr ago (95%
credible interval, 2.8 – 2.1 Myr; figure 3). The coalescent
times of sympatric populations were 0.15 –2.0 Myr, indicating that morphological divergence occurred at various
times during the past 2.4 Myr.
(b) Factors of morphological variation
Both body size and genital size showed close
matching between sexes (figure 4a, r ¼ 0.954; figure 4b,
r ¼ 0.930; p , 0.0001 for both). Genital size was correlated with body size (figure 4c; r ¼ 0.835, p , 0.0001
for males; females: r ¼ 0.890, p , 0.0001); note that
genital size varied more than body size and that differences in body and genital sizes between sympatric
species were highly variable. The matching of genital
sizes between sexes remained even after removing the
effect of body size (r ¼ 0.826, p , 0.0001). The bodyor genital-size distribution did not differ between species
in sympatry and those in allopatry (figure S2a,b in the
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
Multiple speciation events
P33–34
P57–58
P28–29
P59–60
P38–39
P42–43
P17–18
P55–56
root node of
tonominea
species
complex
P61–62
P44–45
P46–47
P63–64
3.0
2.5
2.0
1.5
1.0
0.5
Y369_P57
Y280_P57
Y39_P33
Y51_P33
Y52_P33
Y281_P58
Y42_P34
Y41_P34
Y91_P29
Y284_P59
Y81_P38
Y89_P28
Y327_P28
Y365_P55
Y285_P60
Y88_P18
Y84_P39
Y58_P42
Y288_P63
Y60_P43
Y85_P17
Y87_P17
Y86_P17
Y368_P56
Y275_P56
Y65_P45
Y66_P45
Y49_P45
Y358_P45
Y376_P61
Y379_P61
Y287_P62
Y37_P47
Y63_P47
Y286_P61
Y48_P44
Y47_P44
Y64_P44
Y355_P44
Y36_P46
Y289_P64
Y105_P77
Y106_P7
Y10 doenitzi
Y43 crenata
Y4 laminata
Y2 longa
Y11 ishiii
Y9 erythrosoma
0
divergence time (Myr)
Figure 3. Divergence times between populations of the
tonominea species complex. For the nodes of the most
recent common ancestors of two sympatric species, the
95% credible range of the estimated ages is indicated by
grey bars.
electronic supplementary material), with equivalent levels
of variation among populations. The size difference
between sympatric species was not biased toward large
values against the difference between random population
pairs (figure S2c,d in the electronic supplementary
material) and was not correlated with genetic distance
(figure S2e, f in the electronic supplementary material).
In the multiple regression analysis for environmental
correlates of body- and genital-size variation, this variation was not explained by any of the five geographical
or climatic parameters, although the effects of annual
rainfall on male body size, male genital size and female
genital size were marginally significant (table 1). In
addition, the body-size difference between populations
was not correlated with geographical or genetic distance
between populations (table 2). Similarly, the genital-size
difference between populations could not be explained
by geographical or genetic distance, whereas it was clearly
correlated with the genital size of the opposite sex and
weakly with body size (table 2).
Finally, the evolutionary rates and correlated evolution
of body and genital sizes were analysed using phylogenetically independent contrasts. The absolute value of
the standardized phylogenetically independent contrast
(n ¼ 35) was 1.5 times larger for genital size than body
Proc. R. Soc. B (2010)
T. Sota & T. Tanabe
693
size, suggesting faster evolution of genital size (table 3).
The contrasts of body and genital sizes showed correlated
evolution between sexes (body size: r ¼ 0.811, p ,
0.0001; genital size: r ¼ 0.886; p , 0.0001). Although
the evolution of genital size was correlated with that of
body size in each sex (table 3), the residuals of the genital-size contrast regressed on body size were correlated
between the sexes (figure 4d; r ¼ 0.729; p , 0.0001),
indicating that the correlated evolution in genital size
was partly independent of that in body-size evolution.
4. DISCUSSION
Populations of the tonominea species complex have
diverged extensively in key mating traits, body size and
genital size. Notably, no significant correlation was
observed between the differences in these sizes and the
geographical/genetic distances between populations
(table 2), despite the fact that the genetic divergence
between populations as revealed by COI–COII and
EF-1a sequences showed clear isolation-by-distance patterns, implying gradual dispersal of the millipedes
(figure 2). In addition, climatic variables did not explain
the body-size variation (table 1), suggesting that climatic
adaptation (Masaki 1967; Roff 1980) is not the cause of
the body-size variation. The climatic variables used were
rather coarse and might not reflect soil microhabitat conditions appropriately. For example, the marginally
significant effect of annual rainfall on body size
(table 1) may imply an effect of soil moisture. However,
the sympatry of different-sized millipedes (actually
within the same plots) suggests that local environmental
variables are not major determinants of the body size variation. Nevertheless, our study has not resolved the causes
of body-size variation. A proximate cause of body size
variation may be variation in the larval developmental
period, which may require several years in Parafontaria
(Fujiyama 1996), but the life cycle of the tonominea
species complex is not well known. Obviously, detailed
life history and population studies of sympatric species
pairs are needed.
Despite the above reservation, the lack of phylogenetic,
geographical and climatic correlates with body size,
as well as the sympatry of species pairs with diverged
body sizes, suggest that body size has diverged rapidly,
primarily in response to intrinsic factors within populations, even though the range of body sizes has been
constrained by natural selection. Similarly, although genital sizes show positive correlations with body size, the
genital-size variation has occurred partly independently
of body-size variation (figure 4c,d ). Our previous study
revealed that size matching between the sexes in both
body and genital size is required to accomplish copulation
and that complex interactions take place between male
and female genitalia during insemination (Tanabe &
Sota 2008), which suggests that sexual selection could
be the primary process leading to morphological divergence in the tonominea species complex. The role of
reinforcement or reproductive character displacement in
morphological divergence has not been demonstrated
because the degree of morphological difference between
sympatric species was as variable as that between
random pairs of species, similar to the conclusions of a
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
694 T. Sota & T. Tanabe Multiple speciation events
(a) 1.00
(b) 0.35
0.30
0.25
female genital size
female body size
0.95
0.90
0.85
0.20
0.15
0.10
0.05
0.80
0
0.75
0.75
0.80
0.85 0.90 0.95
male body size
residual female genital size contrast
0.3
male genital size
0
0.1
0.2
0.3
male genital size
0.4
(d ) 0.04
(c) 0.4
0.2
0.1
0
–0.1
0.75
–0.05
–0.1
1.00
0.03
0.02
0.01
0
–0.01
–0.02
0.80
0.85 0.90 0.95
male body size
1.00
0
0.01
0.02
0.03
0.04
residual male genital size contrast
Figure 4. (a) Relationship between population mean body sizes (tenth metatergal width) of males and females of the tonominea
species complex. (b) Relationship between population mean male and female genital sizes (dry weight). (c) Relationship
between population mean male body size and genital size. The sympatric species are connected by grey lines. (d) Relationship
between residuals of phylogenetically independent contrasts of male and female genital sizes, which are regressed on the
phylogenetically independent contrasts of body sizes.
Table 1. Multiple regression of body and genital sizes on three geographical and two climatic variables. Because no
significant model was found with the stepwise regression, the effects (t with p-value in parentheses) in models with all
variables are given.
variable
male body size
(n ¼ 58)
male genital size
(n ¼ 55)
female body size
(n ¼ 47)
female genital size
(n ¼ 43)
latitude
longitude
altitude
annual mean temperature
annual rainfall
20.16 (0.875)
0.65 (0.519)
20.48 (0.633)
20.14 (0.893)
1.97 (0.054)
20.22 (0.824)
1.01 (0.317)
20.29 (0.772)
20.03 (0.978)
1.82 (0.075)
0.34 (0.737)
0.09 (0.927)
0.16 (0.875)
0.40 (0.688)
1.38 (0.175)
0.16
0.52
0.30
0.38
1.95
previous analysis using multidimensional shape
parameters (Tanabe et al. 2001).
Our observation that matching between the sexes in
body size and in genital size is required, and hence probably selected for achievement of copulation (Tanabe &
Sota 2008), implies that correlated evolution between
the sexes has occurred in response to intersexual selection. Common genetic regulation and common natural
selection on sex-limited characters also can affect correlated evolution (Arnqvist & Rowe 2002). Underlying
common genetic regulation is invoked not only for body
Proc. R. Soc. B (2010)
(0.874)
(0.608)
(0.766)
(0.704)
(0.059)
size but also for genital size. However, the male and
female genitalia of millipedes are not homologous
organs (Hopkin & Read 1992); in Parafontaria, the male
gonopods are transformed eighth legs, whereas the
female genitalia comprise a complex of modified genital
openings, adjacent legs and a sternite (Tanabe et al.
2001). Therefore, their simultaneous evolution is not
necessarily governed by a single genetic factor. In
addition, genital size has evolved independently of body
size to an appreciable extent (figure 4d ), suggesting that
body size only roughly constrains genital size. Matching
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
Multiple speciation events
T. Sota & T. Tanabe
695
Table 2. Matrix correspondence tests for variation in (a) body size and (b) genital size among populations. In (a),
geographical and genetic distance (mitochondrial gene sequence) are regressed on Euclidean distances of body size between
populations for each sex. In (b), Euclidean distances of body size of one’s own sex and genital size of the opposite sex,
geographical distance and genetic distance are regressed on the Euclidean distance of genital size between populations for
each sex.
dependent variable
variable
(a) body size
geographical distance
genetic distance
(b) genital size
body size of own sex
genital size of
opposite sex
geographical distance
genetic distance
t
p
t
p
male
4.21
21.21
female
0.141
0.578
2.78
1.16
3.57
27.03
0.068
0.000
16.66
28.75
0.000
0.000
24.00
0.47
0.100
0.801
0.96
20.49
0.681
0.787
male
0.300
0.537
female
Table 3. Evolutionary rates and correlations of body and genital sizes based on phylogenetically independent contrasts.
Standardized contrasts with the same letter do not differ significantly by Wilcoxon sign test (after sequential Bonferroni
correction).
regression of genital size on body size
character
absolute value of standardized
contrast (mean + s.e.)
male body size
male genital size
female body size
female genital size
0.0140 + 0.00176bc
0.0233 + 0.00321a
0.0136 + 0.00159c
0.0196 + 0.00257ab
of body size between the sexes is required in a different
phase of copulation than matching of genital size (i.e.
body alignment and intromission), and hence body size
and genital size may be subject to different kinds of selection. For example, at intromission, the interplay between
genitalia is complex, possibly involving tactile communication and mechanical interaction between sexes
(Eberhard 1985; Tanabe & Sota 2008), and can be
subject to an arms race towards exaggeration. Thus,
divergent evolution, probably driven by sexual selection,
in millipedes can result in various combinations of body
sizes and genital sizes between once-diverged populations, and thus provides great potential to produce
mechanically isolated species in morphological space.
The estimated divergence time of mitochondrial haplotypes from sympatric species ranged from 0.15 to
2.0 Myr, suggesting large variation in the divergence
time. On the mitochondrial gene tree (figure 3), the
time to the bifurcation leading to sympatric species was
relatively short, 0.03– 0.39 Myr (mean, 0.15 + 0.12 s.d.
Myr). Because our population sampling is far from complete, many lineages remain unsampled and hence the
actual divergence of populations could be very rapid.
For species pairs with longer divergence times,
pre-zygotic isolation other than mechanical isolation
and post-zygotic isolation may have evolved, as total isolation of 0.90 is attained in 1.1 Myr on average between
allopatric Drosophila species (Coyne & Orr 2004). In
our previous study on reproductive isolation (Tanabe &
Sota 2008), we used a sympatric species pair with
Proc. R. Soc. B (2010)
slope
t
p
1.403 + 0.171
8.21
,0.0001
1.217 + 0.150
8.12
,0.0001
relatively large morphological differences (figure 1b;
P33 –34) and this pair had the second smallest genetic
divergence (0.21 Myr; figure 3). The species pair did
not discriminate against heterospecific mates and hybridization was hindered by mechanical barriers during
copulation. This case demonstrates that substantial morphological divergence preceded divergence in mate
recognition traits (e.g. sexual pheromones). However,
the inability to discriminate heterospecific mates can
incur direct costs (such as injury) when copulatory
organs are mismatched between sexes (e.g. Sota &
Kubota 1998). Therefore, species pairs with longer divergence times may have acquired pre-mating isolation
mechanisms through divergence of some signal traits.
Explosive radiation of species within a lineage is known
in Hawaiian Drosophila and African lake cichlids (Ritchie
2007). However, these examples of radiation are associated with marked divergence in ecological characters,
particularly in the use of microhabitats and food. Thus,
this millipede case is remarkable, although the animals
are inconspicuous, because their radiation is based on
morphological divergence in agents of mechanical reproductive isolation without marked sign of ecological
divergence. Similar cases of explosive but cryptic speciation are likely in various arthropod groups with internal
fertilization and low dispersal power, and such groups
may be beyond the resolution of morphology-based taxonomy and escape the notice of evolutionary biologists.
Our study suggests that genital and other morphologies
in arthropods that can cause mechanical reproductive
Downloaded from http://rspb.royalsocietypublishing.org/ on May 7, 2017
696 T. Sota & T. Tanabe Multiple speciation events
isolation are sensitive to sexual selection, contribute to
mechanical reproductive isolation (at least before the
evolution of pre-mating isolation) and enable speciation.
We thank Y. Takami for comments and Y. Kawakami, Y. Takai,
T. Wada, N. Tsurusaki, Y. Nishikawa, G. Takaku,
Y. Takashima, K. Yahata, M. Hashimoto, N. Nagata and
H. Nishi for sampling. This study was supported by Grantsin-Aid for scientific research (nos 15207004, 20370011,
21570097) from the Japan Society for the Promotion of
Sciences, and the Global COE programme A06 ‘Formation
of a Strategic Base for Biodiversity and Evolutionary
Research; from Genomics to Ecosystems’ from the Ministry
of Education, Culture, Sports and Technology, Japan.
REFERENCES
Andersson, M. 1994 Sexual selection. Princeton, NJ:
Princeton University Press.
Arnqvist, G. 1998 Comparative evidence for the evolution of
genitalia by sexual selection. Nature 393, 784 –786.
(doi:10.1038/31689)
Arnqvist, G. & Rowe, L. 2002 Correlated evolution of male
and female morphogenesis in water striders. Evolution 56,
936 –947.
Bidau, C. J. & Martı́, D. A. 2008 Geographic and climatic
factors related to a body-size cline in Dichroplus pratensis
Bruner, 1900 (Acrididae, Melanoplinae). J. Orthptera
Res. 17, 149–156. (doi:10.1665/1082-6467-17.2.149)
Casgrain, P. & Legendre, P. 2004 The R-package for
multivariate and spatial analysis v. 4.0 5d-user’s manual.
Department Des Sciences Biologiques, Universite de
Montreal. Available at http://www.fas.umontreal.ca/
BIOL/legendre/.
Coyne, J. A. & Orr, H. A. 2004 Speciation. Sunderland, MA:
Sinauer Associates.
Darwin, C. 1871 The descent of man and selection in relation to sex.
London, UK: J. Murray.
Eberhard, W. G. 1985 Sexual selection and animal genitalia.
Cambridge, MA: Harvard University Press.
Excoffier, L., Laval, L. G. & Schneider, S. 2005 ARLEQUIN
v. 3.0: an integrated software package for population
genetics data analysis. Evol. Bioinform. Online 1, 47–50.
Fujiyama, S. 1996 Annual thermoperiod regulating an eightyear life-cycle of a periodical diplopod, Parafontaria
laminata armigera Verhoeff (Diplopoda). Pedobiologia 40,
541 –547.
Felsenstein, J. 1985 Phylogenies and the comparative
method. Am. Nat. 125, 1–15. (doi:10.1086/284325)
Gavrilets, S. 2000 Rapid evolution of reproductive barriers
driven by sexual conflict. Nature 403, 886 –889. (doi:10.
1038/35002564)
Garland, T., Harvey, P. H. & Ives, A. R. 1992 Procedures for
the analysis of comparative data using phylogenetically
independent contrasts. Syst. Biol. 41, 18–32. (doi:10.
2307/2992503)
Hendry, A. P. 2009 Speciation. Nature 458, 162 –164.
(doi:10.1038/458162a)
Ho, S. Y. W., Phillips, M. J., Cooper, A. & Drummond, A. J.
2005 Time dependency of molecular rate estimates and
systematic overestimation of recent divergence times.
Mol. Biol. Evol. 22, 1561–1568. (doi:10.1093/molbev/
msi145)
Hopkin, S. P. & Read, H. J. 1992 The biology of millipedes.
New York, NY: Oxford University Press.
Huelsenbeck, J. P. & Ronquist, F. 2001 MRBAYES: Bayesian
inference of phylogenetic trees. Bioinformatics 17,
754 –755. (doi:10.1093/bioinformatics/17.8.754)
Proc. R. Soc. B (2010)
Iwasa, Y. & Pomiankowski, A. 1995 Continual change in mate
preference. Nature 377, 420–422. (doi:10.1038/377420a0)
Lande, R. 1982 Rapid origin of sexual isolation and character
divergence in a cline. Evolution 36, 213 –223. (doi:10.
2307/2408039)
Maddison, W. P. & Maddison, D. R. 2004 MESQUITE:
a modular system for evolutionary analysis, v. 2.6.
Available at http://mesquiteproject.org.
Manly, B. F. J. 1986 Randomization and regression methods
for testing for associations with geographical,
environmental and biological distances between
populations. Res. Popul. Ecol. 28, 201–208. (doi:10.
1007/BF02515450)
Manly, B. F. J. 1997 RT: a program or randomization testing.
Cheyenne, WY: West, Inc.
Masaki, S. 1967 Geographic variation and climatic
adaptation in a field cricket (Orthoptera: Gryllidae).
Evolution 21, 725– 741. (doi:10.2307/2406770)
Midford, P. E., Garland, T. & Maddison, W. P. 2003. PDAP
package. http://mesquiteproject.org/pdap_mesquite.
Mousseau, T. A. 1997 Ectotherms follow the converse to
Bergmann’s rule. Evolution 51, 630 –632. (doi:10.2307/
2411138)
Panhuis, T. M., Butlin, R., Zuk, M. & Tregenza, T. 2001
Sexual selection and speciation. Trends Ecol. Evol. 16,
364 –371. (doi:10.1016/S0169-5347(01)02160-7)
Ritchie, M. G. 2007 Sexual selection and speciation. Annu.
Rev. Ecol. Evol. Syst. 38, 79–102. (doi:10.1146/annurev.
ecolsys.38.091206.095733)
Roff, D. 1980 Optimizing development time in a seasonal
environment: the ‘ups and downs’ of clinal variation.
Oecologia 45, 202 –208. (doi:10.1007/BF00346461)
Shapiro, A. M. & Porter, A. H. 1989 The lock-and-key
hypothesis: evolutionary and biosystematic interpretation
of insect genitalia. Annu. Rev. Entomol. 34, 231 –245.
(doi:10.1146/annurev.en.34.010189.001311)
Sota, T. & Hayashi, M. 2007 Comparative historical biogeography of Plateumaris leaf beetles (Coleoptera:
Chrysomelidae) in Japan: interplay between fossil and
molecular data. J. Biogeogr. 34, 977 –993. (doi:10.1111/
j.1365-2699.2006.01672.x)
Sota, T. & Kubota, K. 1998 Genital lock-and-key as a selective agent against hybridization. Evolution 52, 1507– 1513.
(doi:10.2307/2411321)
Sota, T., Takami, Y., Kubota, K. & Ishikawa, R. 2000
Geographic variation in the body size of some Japanese
Leptocarabus species (Coleoptera, Carabidae): The
‘toppled-domino pattern’ in species along a geographic
cline. Entomol. Sci. 3, 309 –320.
Takami, Y. & Sota, T. 2007 Rapid diversification of male genitalia and mating strategies in Ohomopterus ground beetles.
J. Evol. Biol. 20, 1385–1395. (doi:10.1111/j.1420-9101.
2007.01338.x)
Tanabe, T. & Sota, T. 2008 Complex copulatory behavior
and the proximate effect of genital and body size differences on mechanical reproductive isolation in the
millipede genus. Parafontaria. Am. Nat. 171, 692 –699.
(doi:10.1086/587075)
Tanabe, T., Katakura, H. & Mawatari, S. F. 2001
Morphological difference and reproductive isolation:
morphometrics in the millipede Parafontaria tonominea
and its allied forms. Biol. J. Linn. Soc. 72, 249–264.
Thorne, J. L. & Kishino, H. 2002 Divergence time and evolutionary rate estimation with molecular data. Syst. Biol.
51, 689–702. (doi:10.1080/10635150290102456)
West-Eberhard, M. J. 1983 Sexual selection, social competition, and speciation. Quart. Rev. Biol. 58, 155 –183.
(doi:10.1086/413215)