Download Genetics of a wing size difference between two Nasonia species

Genetics of a wing size difference between two Nasonia species
R. F. WESTON, I. QURESHI & J. H. WERREN
Biology Department, University of Rochester, Rochester, NY 14627, USA
Keywords:
Abstract
evolutionary genetics;
Nasonia;
parasitic wasps;
speciation;
wing size.
Very little is known about the genetics of morphological differences between
species. This study investigates the genetic basis of a signi®cant morphological
difference between males of two closely related species of the parasitoid wasp
Nasonia. One of the de®ning characters of species in the genus Nasonia is male
forewing size. The forewings of Nasonia giraulti males are 2.4 times larger than
the forewings of Nasonia vitripennis males. Genetic analysis of hybrids between
these species indicates that this difference is due to the effect of a few genes.
Also discussed is the possible role of `pseudo linkage' in analysis of F2 hybrids.
Pseudo linkage occurs when genes affecting a trait are linked to interacting
hybrid lethal loci, and can lead to an overestimation of the number of regions
involved in a phenotype. The large wing trait of N. giraulti was introgressed
into a N. vitripennis background. Analysis of this introgression line indicates
that 44% of the difference in wing size between the species is due to the
presence of a single gene, or a few tightly linked genes, located on linkage
group IV. Furthermore, the introgressed region appears to affect the width of
the wing more strongly than the length. Indirect results suggest that this
region affects wing cell size, rather than cell number. Results are consistent
with the view that morphological and adaptive differences between species
can have a simple genetic basis.
Introduction
Over the past century there have been various studies on
the genetic basis of morphological differences within and
between species (see reviews by Lande, 1981; Orr &
Coyne, 1992). These studies have examined the genetic
architecture of morphological traits in domesticated
strains of plants and animals and their wild relatives,
strains that differ because of arti®cial selection in the
laboratory; and species or subspecies that differ morphologically, presumably because of differing natural and
sexual selective pressures. Mutations of large effect are
common in animal and plant breeding (Gottlieb, 1984;
Piper & Shrimpton, 1989) but arti®cial selection experiments in the laboratory have resulted in responses due
to mutations of both large effect (Crow, 1957) and small
Correspondence: John H. Werren, Biology Department, University of
Rochester, Rochester, NY 14627, USA.
Tel: 716 275 3889; fax: 716 275 2070;
e-mail: werr@uhura.cc.rochester.edu
586
effect (Crow, 1957; Lande, 1981; Shrimpton & Robertson, 1988). In disturbed natural populations, however,
adaptations are commonly based on mutations with large
effects (Lees, 1981; McNair, 1981).
Less common are genetic analyses of morphological
differences between natural populations or closely related species. Examples include colour differences in Lepidoptera (Clarke & Sheppard, 1959; Sheppard, 1962;
Turner, 1981) and newts (Spurway, 1953), ¯oral traits in
closely related species of monkey¯owers (Bradshaw
et al., 1995) and morphological differences between
sympatric species of sticklebacks (Hat®eld, 1997). Genes
of large effect have been found in ¯oral character
differences between closely related monkey ¯owers
(Bradshaw et al., 1995) and some morphological characters of stickleback species (Hat®eld, 1997). However,
differences in genital arch morphology between closely
related Drosophila species are due to many genes of small
effects (Coyne et al., 1991; Liu et al., 1996). A dif®culty
in studying morphological differences between closely
related species is that typically the differences are small
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
Genetics of species wing size differences
or subtle, or, in cases of signi®cant morphological
divergence, the species cannot be crossed or lack availability of techniques making them suitable for genetic
analyses.
The purpose of this study is to investigate the genetic
basis of a large wing size difference between males of two
closely related species of the parasitoid wasp Nasonia. The
genus Nasonia consists of three parasitic wasp species: N.
vitripennis (Nv), N. giraulti (Ng) and N. longicornis (Nl)
(Darling & Werren, 1990). N. vitripennis is a cosmopolitan
species while N. longicornis is found only in the western
United States, and N. giraulti is found in the north-east
United States (Darling & Werren, 1990). One of the
primary morphological differences between the species is
male forewing size. N. vitripennis males have short,
narrow vestigial wings and the males are incapable of
¯ight. Ng wings are »1.35 times longer than and 1.89
times wider than Nv. N. longicornis (Nl) males have
intermediate sized wings (»1.2 times longer and 1.5 times
wider than Nv). Preliminary experiments indicate that N.
giraulti and N. longicornis males are capable of limited
¯ight.
The three species will mate with one another and, once
cured of their associated cytoplasmic bacterial infections
(Wolbachia), form viable and fertile hybrids (Breeuwer &
Werren, 1990, 1995). Due to the haplodiploid sex
determination in Nasonia, F1 hybrids produced in interspeci®c crosses are female, only males are derived from
unfertilized eggs and therefore have the genotype of the
maternal species. Virgin F1 hybrid females produce all
male (haploid) progeny. While partial F2 hybrid breakdown occurs, survival rates are suf®cient to allow
backcrossing and introgression of genetic regions between the species. The ability to perform interspeci®c
crosses within Nasonia, availability of genetic markers
and existence of haplodiploid sex determination facilitates our analysis of the genetic basis of forewing size
differences between the species. By analysing F2 recombinant hybrid males from these crosses we can determine
whether the Nasonia male wing size difference is due to
many genes of minor effect or to a few genes of large
effect. In addition, we are able to introgress genes
involved in the wing size difference from one species
into the genetic background of the other. These techniques have allowed us to determine that »44% of the
difference in the average normalized wing multiples
between the species is due to the presence of a single
gene, or a region containing tightly linked genes. Additionally, it appears that the region affects wing width
more strongly than wing length, causing a 26% increase
in wing width but only a 17% increase in wing length.
Materials and methods
The general biology of Nasonia has been described by
Whiting (1967). Nasonia are ectoparasitoids of a number
of calliphorid ¯ies found in bird nests and carcasses. All
587
experiments were performed at 25 °C on Sarcophaga
bullata with constant light. Under these conditions the
generation time of Nasonia is »14 days.
Nasonia stocks
A number of strains antibiotically cured of cytoplasmic
incompatibility micro-organisms (Wolbachia) were used
for these experiments. Antibiotically cured strains are
used because elimination of the Wolbachia permits
formation of F1 hybrids (Breeuwer & Werren, 1990).
AsymC is an antibiotically cured strain of the N.
vitripennis wild-type strain (LabII). For the purpose of
this paper AsymC is designated as Nv. Similarly, an
antibiotically cured strain of N. giraulti (RV2T) is designated as Ng[G], where the G in brackets indicates the
strain has a N. giraulti cytotype. The cytotype refers to
heritable cytoplasmic factors, such as mitochondria,
found in the cytoplasm of an individual. A third strain
(R16A) is an introgression stock, created by 16 generations of backcrossing, of the N. giraulti nuclear complement into a N. vitripennis cytotype (see Breeuwer &
Werren, 1995, for a description). This strain is designated
Ng[V]. N. vitripennis has ®ve chromosomes and is characterized by ®ve linkage groups (Whiting, 1967). In
addition, ®ve antibiotically cured N. vitripennis mutant
eye strains (corresponding to the ®ve linkage groups)
were used to determine linkage of the N. giraulti wing
type: RED833R (linkage group I), RDH5R (linkage group
II), BK424tet (linkage group III), OR123R (linkage group
IV) and ST318tet (linkage group V).
Wing measurements
Forewings were removed from the body of male wasps
and both bodies and wings were mounted on a slide
covered with double-sided cellophane tape. Measurements were taken using a dissecting scope at 25´
magni®cation and ocular micrometer. The length of
the wing is de®ned as the distance from its distal end to
the mass of dark connective tissue at the proximal. Only
the transparent portion of the wing was included in the
length measurement. The width of the wing was measured at its widest part, perpendicular to the length
measurement. In addition to the wing measurements,
the width of the head between the eyes in the region just
anterior to the ocelli was measured (referred to as head
width). Head width has been shown to be a good
approximation for relative body size and fecundity
(Skinner, 1983).
The bristle density in a region of the wing was also
measured. Bristle density was approximated by counting
the number of bristles in a 0.0156-mm2 grid. This grid
was positioned just below the end of the stigmal vein
and every bristle with its base within the grid was
counted, including those on the underlying side of the
wing.
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
588
R . F. W E S T O N E T A L .
Transformations
We used a `normalized wing multiple' to characterize
wing sizes. The `wing multiple' was calculated by
multiplying the width of a wing by its length. However,
wing size increases allometrically with body size in both
species (Nv P < 0.001, Ng P < 0.001, Linear Regression).
Therefore, the wing multiple was normalized for body
size by dividing it by the measured head width of each
wasp. The normalized wing multiple eliminates most of
the correlation between wing size and body size in both
species (n ˆ 75, P > 0.50 for Nv, n ˆ 113, P > 0.5 for
Ng[V], Linear Regression).
Accuracy in measurements
To check the accuracy and reproducibility of the wing
measurements, measurements of the Nv parental strain
were repeated a month after their initial measuring.
These measurements were performed without any
knowledge of the original measurements. Comparison
of the second measurement to the ®rst showed an
average 0.4 ‹ 1.8SD difference in wing multiples
(n ˆ 49). This difference is small compared to the mean
normalized wing multiple of Nv (20.1 ‹ 2.1SD, n ˆ 49,
1.9% difference) or to the difference between Nv and Ng
(29, 1.3% difference).
F2 hybrid crosses
For all interspeci®c crosses, males and females were
collected as virgin pupae and allowed to eclose. After
eclosion, females were mass mated to males (3
males ´ 10 females) or left as virgins. Because Nasonia
are haplodiploid, virgin females produce only male
progeny by parthenogenesis. After 24 h, males were
removed (in mated crosses), and females were transferred to individual vials and provided with two hosts.
Females were transferred to fresh hosts after 2 days. To
produce F2 hybrid strains for linkage analysis, males of all
the N. vitripennis strains (AsymC, RED833R, RDH5R,
BK424tet, OR123R, ST318tet) were crossed to Ng[V]
females. After 11 days, hosts were opened and wasp
pupae sexed. The F1 female hybrids were collected as
virgins and allowed to eclose. Adult F1 hybrid virgins
were set singly on two hosts to obtain F2 recombinant
hybrid males, who were scored for wing phenotype.
Introgression line (INTW1)
The INTW1 stock was created by crossing N. vitripennis
females to N. giraulti males, and then backcrossing F1
females to N. vitripennis males. In each subsequent
generation, virgin females from families also containing
some large-winged males were crossed to N. vitripennis,
thus selecting for the Ng large-wing trait while introgressing into an Nv background. These backcrosses were
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
carried out for ®ve generations, in principle reducing the
N. giraulti genome to 3.1% of the total genome. This
resulted in the INTW1 line which is almost completely N.
vitripennis in genome, contains Nv cytotype, and has a
male wing size intermediate to Nv and Ng. After ®ve
generations of introgression, the INTW1 strain was
maintained heterozygously for several generations, then
large winged males were crossed to Nv females and the
progeny were maintained into diapause. These were
removed from diapause in April 1996 and purebred for
the large wing trait.
Purebred Wg lines
After removal from diapause, virgin INTW1 females were
mated to Nv males. To produce a pure-bred large winged
line, virgin daughters of mothers producing large winged
sons were mated to large winged males. This was
repeated each generation until we had produced families
homozygous for the Ng large wing trait (Wg). During the
initial crosses of Wg/Wg females to Wg males, ®ve
isofemale sublines were created. A single line, IWTW1.1,
was used for subsequent studies. To determine the
linkage of the large-wing trait, INTW1.1 males were
mated singly to mutant females of each eye marker.
Males were mated to a mutant female and then transferred to a female carrying a different mutant marker.
These females were individually given two hosts and
their F1 female pupae were collected and allowed to
eclose. Adult F1 virgin females were provided with two
hosts and their male progeny were scored for segregation
of the N. giraulti wing trait.
Estimating the number of genes involved in wing size
The minimum number of effective factors was calculated
using the Wright±Lande equation adapted for a haplodiploid system (eqn 1) (Orr and Jones, personal communication). Where ne is the minimum number of effective
loci involved in the trait, P1 and P2 are the parental
means for trait A, and rA is the variance in trait A among
the progeny. This provided an estimate of the minimum
number of genes involved in wing size determination
between Nv and Ng:
ne ˆ …P2 ÿ P1 †2=4 r2A :
…1†
Statistics
Student's t-tests were used to determine whether differences in wing, head and bristle measurements between
strains were statistically signi®cant. For these tests a
comparison of two means was used in order to determine
the probability that the mean normalized wing multiples
of the mutant and wild-type F2 males were different from
one another. To determine whether two nonparametric
distributions differed in their dispersion around the
Genetics of species wing size differences
589
median, a Mann±Whitney U-test was performed on the
absolute values of the difference between the medians
and observed data. We refer to this test as Deviation
From the Median (DFM).
Results
Characterization of Nv and Ng species wing size
differences
Measurements of length, width and head size of the wild
type strains of Nv and Ng[G] were performed. Head size
does not vary signi®cantly between the species (Nv
9.2 ‹ 0.8SD, n ˆ 75, Ng 9.4 ‹ 0.7SD, n ˆ 83) (P > 0.05).
Analysis of wings revealed wing lengths between the
species overlap slightly, but wing widths do not overlap.
The average measurements are provided in Table 1. Both
length (P < 0.001) and width (P < 0.001) differ signi®cantly between the species. The same results were found
after normalizing the length and width measurements
with respect to overall wasp body size. After transforming
the measurements to a normalized wing multiple, which
is a rough estimate of wing area normalized to body size,
the N. vitripennis (Nv) and N. giraulti (Ng) distributions
proved to be completely nonoverlapping (Fig. 3) and
mean values are statistically different (P < 0.001 Fig. 1).
Additionally the wild type Ng[G] was compared to the
Ng[V] strain, which possesses a N. giraulti nuclear
genome and a N. vitripennis cytotype. No signi®cant
differences were found between male wing size in these
strains.
Previous work in Drosophila shows that the number of
wing bristles can provide an estimate of the number of
cells in the wings (Dobzhansky, 1929). If a similar pattern
occurs in Nasonia (not yet established) the bristle density
could be used to estimate cell size in wings. The wing
bristle density characteristic of each species was determined. Bristle density in Nv (133 ‹ 22SD, n ˆ 48) is
approximately three times higher than in Ng[V]
(46 ‹ 7SD, n ˆ 75) (P < 0.01). The lower bristle density
suggests N. giraulti has larger wing cells than in
N. vitripennis (Fig. 2).
Table 1 Basic characterization
of the strains. The mean and
standard deviation of each
measurement is given. Measurements are in ocular units
(1 unit = 0.2 mm).
Fig. 1 The distribution of normalized wing multiples in Nv ´ Ng[V]
F2 males. (a) The normalized wing multiples of male forewings
from the Nv and Ng[V] parental strains. (b) The initial cross of a Nv
male to a Ng[V] female shows a nonunimodal distribution.
F2 hybrids
If the difference in male wing size between N. vitripennis
and N. giraulti is due to the presence of many genes of
small effect, a unimodal and normal distribution of wing
size is expected among the F2 males. Alternatively, if the
difference is due primarily to a gene of large effect, a
bimodal distribution is expected as the gene causing a
major phenotypic effect is received by only half of the
progeny. Figure 1 shows the distribution of normalized
Strain
Wing
width
Wing
length
Parentals
Nv
N1
Ng[G]
Ng[V]
7.3 ‹ 0.7
10.9 ‹ 0.6
13.9 ‹ 1.4
14.0 ‹ 0.9
25.3
29.6
34.2
34.5
Introgression lines
INTW1.1
INTW1.2
INTW1.3
10.0 ‹ 0.8
10.0 ‹ 0.5
9.6 ‹ 0.6
30.3 ‹ 1.7
30.3 ‹ 1.4
29.8 ‹ 3.8
‹
‹
‹
‹
Head
width
2.0
1.5
2.9
2.3
9.2
9.4
9.4
9.9
‹
‹
‹
‹
0.8
0.5
0.7
0.7
9.2 ‹ 0.7
9.2 ‹ 0.4
9.0 ‹ 0.5
Normalized
wing
multiple
n
20.1
34.3
50.5
49.1
2.3
2.5
6.1
4.5
75
25
83
113
32.9 ‹ 3.1
33.2 ‹ 2.8
31.8 ‹ 4.6
105
40
40
‹
‹
‹
‹
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
590
R . F. W E S T O N E T A L .
Table 2 Linkage analysis of the Wg wing trait. The mean normalized wing multiple and standard deviation of the F2 males resulting
from all crosses are shown (male/female). Signi®cance is based on
the results of Student's t-tests comparing the mean of normalized
wing multiples of mutant F2 males to the mean of normalized wing
multiples among wild-type F2 males. Measurements are in ocular
units (1 unit = 0.2 mm).
Fig. 2 The number of bristles in a region of the wing next to the
stigmal vein. Bristle number was collected for the INTW1 and
parental lines and plotted against their normalized wing multiples.
Bristle density clearly decreases as wing size increases.
wing multiples among the recombinant F2 males produced when a Nv male is crossed to a Ng[V] female
(n ˆ 124). The distribution is clearly not normal
(v2 ˆ 18.2, P < 0.001; d.f. 6) and appears to be bimodal,
suggesting that there is a major gene affecting wing size.
The dispersion observed in the data around the median
(DFM) is lower in the F2 hybrids than in the pooled
parental population (P < 0.001). The inward shifting of
the distribution indicates either the presence of small
effect modi®ers as well as the presence of genes of major
effect, or alternatively maternal effects of the F1 hybrid
genotype. Using the modi®ed Wright±Lande equation
(see Methods) we calculate the minimum number of
genes involved in male wing size to be 1.6. This
calculation assumes additivity of gene effects. Epistatic
interactions, which can also generate bimodal distributions, could invalidate this estimate.
To further investigate the genetics of the wing size
difference, we performed a linkage analysis of the wing
trait by crossing strains of N. vitripennis carrying various
mutant eye markers, to N. giraulti. This allows us to
determine which linkage group has the largest effect on
wing size. Analysis of the F2 recombinant hybrid males of
these crosses (Table 2) shows the N. vitripennis wing type
segregates strongly with the mutant eye marker OR123,
located on linkage group IV (OR eye F2 males
25.2 ‹ 6.3SD, n ˆ 62; WT eye F2 males 36.7 ‹ 6.4SD,
n ˆ 62). The N. vitripennis wing also shows strong segregation with RED833 (linkage group I) and weaker segregation with ST318 (linkage group V) but not with RDH5
(linkage group II) or BK424 (linkage group III) (Table 2).
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
Mutant
Wild-type
Overall
Parentals
Nv
Ng[V]
INTW1.1
±
±
±
±
±
±
20.1 ‹ 2.3
49.1 ‹ 4.5
32.9 ‹ 3.1
F2 hybrid crosses
Nv/Ng[V]
RED833R/Ng[V]
RDH5R/Ng[V]
BK424tet/Ng[V]
OR123R/Ng[V]
ST318tet/Ng[V]
±
25.1
30.5
30.4
25.2
26.0
‹
‹
‹
‹
‹
7.8
7.8
7.9
6.3
6.5
±
34.2
30.2
30.8
36.2
33.4
‹
‹
‹
‹
‹
8.8
7.5
8.9
7.1
8.4
30.3
29.7
30.3
30.6
30.7
29.5
‹
‹
‹
‹
‹
‹
7.4
9.5
7.6
8.2
8.7
8.3
F2 introgression crosses
INTW1.1/Nv
INTW1.1/RED833R
INTW1.1/RDH5R
INTW1.1/BK424tet
INTW1.1/OR123R
INTW1.1/ST318tet
±
26.4
26.2
26.0
20.3
26.0
‹
‹
‹
‹
‹
7.7
6.7
7.4
2.0
6.5
±
24.7
25.7
26.1
31.9
26.2
‹
‹
‹
‹
‹
6.7
7.3
5.9
3.4
6.2
28.2
25.6
25.9
26.0
26.1
26.1
‹
‹
‹
‹
‹
‹
5.9
7.2
7.1
6.6
6.5
6.4
The linkage analysis of the F2 hybrid males suggests
there may be as many as three regions which produce a
detectable effect on wing size in Nasonia. It is clear a
genetic factor which produces a major effect is tightly
linked to the OR123 locus on linkage group IV. There is
also support for another region on linkage group I and
weaker support for a third on linkage group V. There are
two possible explanations for the observed linkage of the
N. giraulti wing trait to three linkage groups in the F2
hybrid males. There may be more than one gene with
major phenotypic effect involved in the wing size
difference or there may be `pseudo-linkage' between
the N. giraulti wing trait (Wg) on linkage group IV and a
hybrid lethal.
Pseudo-linkage
The effect of negative epistatic interactions in hybrid
backgrounds presents a unique problem in mapping loci
in hybrid crosses. Negative epistasis in hybrids often
results in lethal gene combinations, affecting the results
of linkage analyses because a gene of interest can appear
linked to two linkage groups due to the negative epistatic
interaction. If a gene of interest is tightly linked to a
hybrid lethal, it will also appear to be linked to the region
where the epistatic interactor with that hybrid lethal is
located. This is illustrated in Fig. 3 where linkage of a
trait (i.e. t locus) to a hybrid lethal locus (i.e. lA) creates
the appearance of linkage to a mutant marker. If a
marker, located on a separate linkage group, is linked to
Genetics of species wing size differences
the locus (lB) that interacts with lA the marker then will
also appear linked to the trait locus. In this model the
allele from one species (lA2*) is a recessive lethal when
combined with the interactor allele from the other
species (lB1*). The `*' indicates a lethal allele. Thus, lA2*
lB1* haploid F2 males die. Since the t locus is linked to
one interactor, and the marker is linked to the other,
pseudo-linkage can occur. Given recombination rates of
r1 between lB and m, and r2 between lA and t, the
apparent linkage between m and t is
rpseudo ˆ
2 ÿ r 1 r 2 ÿ … 1 ÿ r 1 † …1 ÿ r 2 †
:
3
…2†
As r2 converges on zero (i.e. tight linkage between the
trait locus and lA), then
rpseudo ˆ
…1 ‡ r 1 †
3
…3†
and similarly, as r1 converges on zero (tight linkage
between the marker and lB) then
rpseudo ˆ
… 1 ‡ r2 †
:
3
…4†
Thus, the maximal pseudo-linkage expected with a
two locus hybrid lethal interaction is
rpseudo ˆ 1=3:
…5†
Note also that pseudo-linkage creates an asymmetry in
the recovery ratio of the two recombinant genotypes
(Fig. 3).
Fig. 3 Diagram of pseudo-linkage to a hybrid lethal in a haplodiploid system. (a) The genotype of the F1 hybrid females: t is the loci of
interest, lA is the ®rst interactor involved in the hybrid lethal, lB is
the second interactor involved in the hybrid lethal; m is a mutant
marker. (b) The probability of different genotypes in the F2 males.
(c) The apparent frequency of recombination between t and m due
to pseudo-linkage.
591
Pseudo-linkage complicates the analysis of phenotypic
traits in interspecies crosses. One approach to resolving
the involvement of particular genetic regions in a
phenotype is to introgress the region into a homogenous
genetic background while eliminating or reducing the
hybrid deleterious effects through recombination.
While N. vitripennis and N. giraulti are capable of
hybridization, some hybrid breakdown occurs. Previous
studies have demonstrated increased mortality among F2
recombinant hybrids due to negative epistatic interactions between nuclear genes (Breeuwer & Werren,
1995). One of the genes involved in the hybrid lethality
(designated la) has been mapped near the OR123 locus
(unpublished data). When males carrying the Ng allele
(lag*) of this gene also carry the Nv interactor (lbv*) the
resulting larvae die. The presence of these hybrid lethals
in the Nasonia system complicates the task of mapping
the Wg wing trait. Since at least one factor involved in
the wing size difference between Nv and Ng is on linkage
group IV, near a hybrid lethal, it will show pseudolinkage
to the second interactor involved in the nuclear±nuclear
lethality.
Introgression lines
The Ng large wing trait was introgressed into a Nv genetic
background by a series of backcrosses. From this line,
three sublines homozygous for the N. giraulti wing trait
were purebred and characterized (Table 1). The three
INTW1 sublines show no signi®cant difference in head
size (P > 0.05) or in normalized wing multiple (P > 0.05)
among the three sublines. The INTW1 sublines have a
mean head width of 9.2 ‹ 0.6 (n ˆ 185), which does not
differ signi®cantly from the Nv mean head width
(9.2 ‹ 0.8, n ˆ 75) (P > 0.05). This indicates INTW1
males are on average the same size as Nv males. In
contrast, INTW1 males have an average normalized wing
multiple of 32.7 ‹ 3.4 (n ˆ 185), which is signi®cantly
larger than the average normalized wing multiple of Nv
males (20.1 ‹ 2.3, n ˆ 75) (P < 0.001). The third species
in the genus, N. longicornis, also has wings intermediate to
those of N. vitripennis and N. giraulti. Nl has an average
normalized wing multiple of 34.3 ‹ 2.5 (n ˆ 25) which
does not differ signi®cantly from the average normalized
wing multiple of the INTW1 sublines (P > 0.05).
Due to the similarities of the IWTW1 sublines, subsequent analysis was performed on one line, IWTW1.1. The
wings of INTW1.1 males are 26% wider and 17% longer
than Nv males. The Wg trait, introgressed into an Nv
background, causes a 63% increase in the average
normalized wing multiple when compared to the wildtype Nv strain. This difference accounts for 44% of the
difference in the average normalized wing multiples
between Nv and Ng (see Fig. 4). The INTW1.1 subline
shows a 16% decrease in bristle density relative to Nv
males (Table 1) and INTW1.1 bristle densities are intermediate to the Nv and Ng densities (Fig. 2). There
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
592
R . F. W E S T O N E T A L .
and may be producing lower quantities of a maternal
product affecting progeny wing size.
Based on the frequency of recombination between
OR123 and the Wg trait (Table 2), we estimate the Wg
locus to be located on linkage group IV, »1.4 map units
from OR123. The results of the linkage analysis indicate
that any loci affecting wing size on linkage groups I and V
were not introgressed into the INTW1 strain, or that
these F2 effects are due to pseudo-linkage.
Discussion
Darwin (1859) ®rst suggested that selection acts on small
differences in phenotypes over time, producing drastically different structures. Darwin's view was contested by
Huxley (1860), who claimed that nature can make
jumps, thus sparking a debate on the importance of large
Fig. 4 The normalized wing multiple of the INTW1 introgression
line (subline INTW1.1). The wings of the INTW1 males are clearly
intermediate to those of the Nv and Ng[G] parental strains.
appears to be no correlation between the bristle density
and the normalized wing multiple within a strain but a
correlation clearly exists between bristle densities and
normalized wing multiples between the Ng, INTW1.1
and Nv strains. Interestingly, while Nl and INTW1.1
males have similar size wings, their bristle densities differ
signi®cantly (Nl 67 ‹ 8, n ˆ 25; INTW1.1112 ‹ 10,
n ˆ 50) (P < 0.001) indicating that the Wg region is not
solely responsible for the intermediate wing phenotype
observed in Nl males.
Linkage analysis of introgression lines
After creation and characterization of the INTW1 sublines (Fig. 4), another linkage analysis was performed to
determine which regions responsible for the observed
wing size difference in F2 recombinant hybrids had been
isolated in the introgression line. The N. giraulti wing trait
introgressed into the INTW1 stock shows strong segregation in the cross to OR123, indicating that we have
indeed introgressed the N. giraulti wing trait (Wg),
responsible for the major difference between the species,
into a Nv background (Table 2). The distribution of
normalized wing multiples among the F2 hybrids, in the
INTW1.1 cross to OR123R, appears bimodal. However,
the distribution is shifted inward relative to the parental
distributions (P < 0.001 Fig. 5). This shift may be due to
the presence of modi®ers of small effect, or the wing
phenotype may be in¯uenced by the expression of the
maternal genotype (i.e. maternal effects). The F1 hybrids
produced in this cross are heterozygous for the Wg trait
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
Fig. 5 Linkage analysis of the Wg trait in the INTW1.1 line. (a) The
normalized wing multiples of the INTW1.1 (n ˆ 105) and OR123
(n ˆ 100) parental strains; (b) the normalized wing multiples of
male forewings found in the F2 males resulting from an INTW1.1
male crossed to an OR123R mutant female (n ˆ 120).
Genetics of species wing size differences
mutations in evolution which has lasted over a century.
Fisher (1958) presented the primary argument against
the evolutionary importance of mutations with large
effect. Fisher claimed large mutations have a very small
chance of being favourable because they tend to radically
alter the function of a gene or eliminate gene function
entirely. Therefore, adaptations must be based on numerous gene substitutions, all of which have relatively
small effects. Fisher's theory claims mutations with major
phenotypic effects on morphology tend to have deleterious pleiotropic effects, mutations with smaller effects
occur more frequently than those with large effects, and
that large effect mutations are more likely to be further
from the adaptive optimum phenotype. Lande (1983) has
developed a quantitative model in support of Fisher's
predictions.
While mutations of large effect can have negative
pleiotropic effects, deleterious pleiotropy may be equally
likely for mutations with minor effects (Orr & Coyne,
1992), and the deleterious pleiotropy of mutations with
major effects can be ameliorated by the presence of
modi®ers (Maynard Smith, 1983). Furthermore, Fisher's
model assumes the population is already near the
selective optimum. Otherwise, mutations with large
effects are not necessarily farther from the optimum
than mutations with small effects. In addition, Kimura
(1983) argued that while mutations of large effect are less
likely to be favourable than mutations of small effect,
mutations with major effects (that move the phenotype
close to the optimum) are ®xed more easily.
An interesting pattern has been observed during the
evolution of insecticide resistance in natural populations
of the blow¯y, Lucilia cuprena (McKenzie et al., 1990;
Clarke, 1997). Initially, selection favoured the increase of
a gene of major effect imparting resistance to organophosphorous insecticides. However, the resistant allele
had negative (pleiotropic) side-effects. Selection subsequently favoured the increase of a modi®er that ameliorated these negative pleiotropic effects. Turning the clock
forward, one might expect the accumulation of additional modi®ers that ameliorate other and/or weaker negative side-effects, as well as modi®ers that re®ne the
insecticide resistance trait. Thus, the genetic architecture
of this trait may be expected to change over time, from
relatively simple with genes of major effect to more
complex (e.g. major gene effects plus many modi®ers).
We expect that this pattern will be common in nature,
particularly when the original phenotype was subject to
strong selection.
In this study we have shown a signi®cant proportion of
a morphological difference between two species of
Nasonia is due to a single gene or a small region of a
few tightly linked genes. Two conclusions can clearly be
drawn from this study. First, the Ng wing trait is not
highly polygenic since it only shows strong linkage to
two linkage groups (I and IV) in F2 hybrid analyses.
Second, there is clearly a major factor located on linkage
593
group IV. Crosses between Nv and Ng result in F2
recombinant hybrid males which show a non-normal,
nonunimodal distribution in forewing size (Fig. 3). The
apparent bimodality of this distribution indicates the
wing size difference between Nv and Ng is primarily due
to a single or few mutations of large phenotypic effect. It
is still not clear how many genes are involved in the Ng
wing trait; however, it is clear that there are major factors
involved.
After determining that a genetic region of large effect is
involved in the species wing size difference, we selected
for large wings while introgressing Ng genes into a Nv
genetic background (creating the INTW1 strain). Analysis
of the INTW1 strain indicates that it carries the Wg large
wing region linked to OR123. Linkage analysis shows
that the INTW1 large wing trait is caused by a gene of
large effect (or a region of tightly linked genes) which is
located on linkage group IV, »2.1 map units from the
OR123 mutant marker. This region of the Ng genome is
responsible for »44% of the difference in average
normalized wing multiple between the Nv and Ng[V]
parental strains. We refer to this gene or region of genes
as Wg. Wg actually causes a 63% increase in the average
normalized wing multiple when compared to the Nv
strain into which it was introgressed. Thus, we have
introgressed one of the major genes involved in Ng large
wings into an Nv nuclear background.
The results of the F2 hybrid analysis indicate that there
is another locus affecting wing size on linkage group I
(Fig. 5), which was not introgressed into INTW1. Since
the N. vitripennis wing trait shows strong segregation with
RED833 this effect is not likely to be caused by pseudolinkage. If there is another locus with a major effect on
wing size located on linkage group I, nearly all of the
variation in wing size between Nv and Ng may be
controlled by these two regions.
It has previously been shown, in Drosophila, that bristle
number in wings is correlated with cell number in wings
(Dobzhansky, 1929). Nv has a bristle density approximately three times that of Ng. Thus, if the trend observed
in Drosophila proves true in Nasonia, then it appears the
wing cells in Ng are larger than in Nv. Additionally the
introgression line, INTW1, has an intermediate bristle
density to the parental lines (Fig. 4).
Wg appears to primarily affect wing width. Wings
show a 26% increase in normalized wing width in the
INTW1 stock, but only a 17% increase in normalized
wing length (Fig. 5). This may be due to increased cell
width or an increase in the number of cells on the width
axis of the wing. The bristle density data seem to support
wider cells, but further studies will have to be conducted
to determine the relationship between bristle number
and cell number in Nasonia wings. Previous research
shows that ploidy effects cell size in a wide range of
species (Cavalier-Smith, 1985). The ploidy of Nv and Ng
wings, and the possibility of alternative mechanisms for
regulating cell size, remain to be investigated.
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
594
R . F. W E S T O N E T A L .
We have created an introgression line (INTW1) which
has wings intermediate in size to those of N. vitripennis
and N. giraulti. N. longicornis also has a wing size
intermediate between Nl and Ng males. The average
normalized wing multiple of Nl males does not differ
signi®cantly from that of INTW1 males, which raises a
question: is the Wg locus responsible for the difference in
wing size observed between N. vitripennis and N. longicornis? Nasonia belongs to the Dibrachys group in the
Pteromalidae, subfamily Pteromalinae. This includes
wasps in the genera Dibrachys, Dibrachoides and
Muscidifurax. Among these genera, vestigial-winged
males have only been found in Nasonia (Wallace,
1973). It therefore seems the ancestral state of the wing
is the large form, and N. vitripennis has evolved smaller
wings either due to mutational degradation or due to
selection for smaller wings. N. giraulti and N. longicornis
are sister species and N. vitripennis is more divergent.
However, all are `young' species, having diverged in the
Pleistocene (0.2±1 Ma) (Campbell et al., 1993).
The adaptive signi®cance of male wing size in Nasonia
remains unclear. The evolution of small wing size in
males could be either an adaptive trait or the result of
mutational degradation of an unused structure. Nevertheless, the loss of long-range male dispersal associated
with wing size reduction is consistent with the mating
population structure of these wasps. All three species of
Nasonia have highly subdivided populations in nature
(Werren, 1983; Darling & Werren, 1990) and show
strongly female-biased sex ratios (Skinner, 1983; Werren,
1983; Drapeau & Werren, 1999), indicative of local
mating (Hamilton, 1967). If mating opportunities for
males are con®ned to the local area where they emerge,
then selection for nondispersing males is expected.
Selection for nondispersing males could result in males
with small vestigial wings. However, this argument
would seem to apply equally to all three species, whereas
small wings evolved in Nv and Nl, but not in Ng.
Preliminary data on male dispersal in Nasonia have
revealed a difference in ¯ight abilities between the
species. Ng males ¯y, but have low ¯ight tendency.
Males of different strains of Nl show variable ¯ight
tendency. Nv males do not ¯y, nor do they exhibit ¯ight
behaviours (i.e. raising wings) (unpublished data). These
results imply that reduction of ¯ight behaviour precedes
the loss of wings.
Evolution of small wings in Nv could have been
selectively favoured due to advantages in male±male
competition. Consistent with this view, Nv males are
more aggressive than are Nl or Ng males, and Nv shows
less female-biased sex ratios, indicative of more mate
competition. In contrast, Ng males and females typically
mate within the host, where inbreeding is likely to be
high and mate competition low (Drapeau & Werren,
1997). Interestingly, Nv males use their vestigial wings in
aggressive displays by raising them up off their backs, a
behaviour not observed in Ng. These species differ in a
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD
number of interesting respects, including courtship and
mating, sex ratio, male aggression, ¯ight behaviour,
morphology (wing size, antennal shape) and host preference. An integrative approach combining behavioural,
genetic and molecular genetic studies promises to be
useful in revealing how these differences have evolved in
the species complex. The ability to move genetic regions
between the species will be particularly useful for such
studies. It will be interesting to determine how often
phenotypic differences between these species are due to
genes of large effect vs. the accumulation of many genes
of small effect.
Acknowledgments
Thanks are extended to Charlotte Lee and Dan Williams
for their contribution to this research, and to C. Arboleda
and B. McAllister for general assistance. Thanks to C.
Brett, S. Bordenstein, V. Calhoun, M. Drapeau, C. Jones,
A. Orr, D. Presgraves and M. Perfeccti, who provided
helpful comments on the manuscript, and R. Hannigan
helped with statistical and graphical programs. C. Jones
and A. Orr kindly provided their derivation for the
minimum number of genes estimate for haplodiploidy.
References
Bradshaw, H.D., Wilbert, S.M., Jr, Otto, K.G. & Schemske, D.W.
1995. Genetic mapping of ¯oral traits associated with reproductive isolation in monkey¯owers (Mimulus). Nature 376:
762±765.
Breeuwer, J.A. & Werren, J.H. 1990. Microorganisms associated
with chromosome destruction and reproductive destruction
and reproductive isolation between two insect species. Nature
346: 558±560.
Breeuwer, J.A. & Werren, J.H. 1995. Hybrid breakdown
between two haplodiploid species: the role of nuclear and
cytoplasmic genes. Evolution 49: 705±717.
Campbell, B.C., Steff, J.D., n-Campell, -e. & Werren, J.H. 1993.
Phylogeny of the Nasonia species complex (Hymenoptera:
Pteromalidae) inferred from an internal transcribed spacer
(ITS2) and 28S rDNA sequences. Ins. Mol. Biol. 2: 225±237.
Cavalier-Smith, T. 1985. Cell volume and the evolution of
eukayotic genome size. In: The Evolution of Genome Size (ed. T.
Calvalier-Smith), pp. 105±184. John Wiley, New York.
Clarke, G.M. 1997. The genetics and molecular basis of developmental stability: The Lucilia story. Trends Ecol. Evolution 12:
89±91.
Clarke, C.A. & Sheppard, P.M. 1960. The evolution of mimicry
in the butter¯y Papilio dardanus. Heredity 14: 163±173.
Coyne, J.A., Rux, J. & David, J.R. 1991. Genetics of morphological differences and hybrid sterility between Drosophila
sechellia and its relatives. Genet. Res. 57: 113±122.
Crow, J.F. 1957. Genetics of insect resistance to chemicals. Ann.
Rev. Ent. 2: 227±246.
Darling, D.C. & Werren, J.H. 1990. Biosystematics of Nasonia
(Hymenoptera: Pteromalidae): two new species reared from
bird's nests in North America. Ann. Ent. Soc. Am. 83: 352±370.
Darwin, C. 1859. On the Origin of Species. John Murray, London.
Genetics of species wing size differences
Dobzhansky, T. 1929. The in¯uence of the quantity and quality
of chromosomal material on the size of the cells in Drosophila
melanogaster. Arch. Entwicklungsmech. Org 115: 363±379.
Drapeau, M.D. & Werren, J.H. 1998. Differences in population
structure and sex ratio between three sibling species of
Nasonia. Evol. Ecol. Res. in press.
Fisher, R.A. 1958. The Genetical Theory of Natural Selection, 2nd
edn. Sinauer, Sunderland, Mass.
Gottlieb, L.D. 1984. Genetics and morphological evolution in
plants. Am. Nat. 123: 681±709.
Hamilton, W.D. 1967. Extraordinary sex ratios. Science 156: 477±
478.
Hat®eld, T. 1997. Genetic divergence in adaptive characters
between sympatric species of stickleback. American Naturalist
149: 1009±1029.
Huxley, T.H. 1860. The origin of species. Westminster Review 17:
541±570.
Kimura, M. 1983. The Neutral Theory of Molecular Evolution.
Cambridge University Press, Cambridge.
Lande, R. 1981. The minimum numbers of genes contributing to
quantitative variation between and within populations.
Genetics 99: 541±553.
Lande, R. 1983. The response to selection on major and minor
mutations affecting a metrical trait. Heredity 50: 47±65.
Lees, D.R. 1981. Industrial melanism: genetic adaptation of
animals to air pollution. In: Genetic Consequences of Man Made
Change (J. A. Bishop & L. M. Cook eds), pp. 129±176.
Academic Press, London.
Liu, J., Merger, J.M., Stam, L.F., Gibson, G.C., Zeng, Z.B. &
Laurie, C.C. 1996. Genetic analysis of a morphological shape
difference in male genitalia of Drosophila simulans and Drosophila mauritiana. Genetics 142: 1129±1145.
Maynard Smith, J. 1983. The genetics of stasis and punctuation.
Annu. Rev. Genet. 17: 11±25.
McKenzie, J.A., Batterham, P. & Baker, M.L. 1990. Fitness and
asymmetry modi®cation as an evolutionary process. A study
in the Australian Sheep Blow¯y, Lucilia cuprena and Drosophila
595
melanogaster. In: Ecological and Evolutionary Genetics of Drosphila
(J. S. S. Barker, W. T. Starmer & R.J. MacIntyre, eds), pp. 57±
73. Plenum Press.
McNair, M. 1981. Tolerance of higher plants to toxic materials.
In: Genetic Consequences of Man Made Change (J. A. Bishop & L.
M. Cook, eds), pp. 177±208. Academic Press, London.
Orr, H.A. & Coyne, J.A. 1992. The genetics of adaptation: a
reassessment. American Naturalist 140: 725±742.
Piper, L.R. & Shrimpton, A.E. 1989. The quantitative effects of
genes which in¯uence metric traits. In: Evolution and Animal
Breeding (W. G. Hill & T. F. C. Mackay, eds), pp. 147±151. C.
A. B. International, Wallingford, UK.
Sheppard, P.M. 1962. Some aspects of the geography, genetics,
and taxonomy of a butter¯y. Tax. Geog. 4: 135±152.
Shrimpton, A.E. & Robertson, A. 1988. The isolation of
polygenic factors controlling bristle score in Drosophila melanogaster. II. Distribution of third chromosome bristle effects
within chromosome sections. Genetics 11: 49±78.
Skinner, S.W. 1983. Extra chromosomal sex ratio factors in the
parasitoid was Nasonia ( ˆ Mormoniella) Vitripennis. PhD thesis,
University of Utah, Salt Lake City, Utah.
Spurway, H. 1953. Genetics of speci®c and subspeci®c differences in European newts. Symp. Soc. Experimental Biol. 7: 200±
237.
Turner, J.R.G. 1981. Adaptation and evolution in Heliconius: a
defense of NeoDarwinism. Ann. Rev. Ecol. Syst 12: 99±121.
Wallace, G.E. 1973. New Pteromalidae of the Dibrachys group
(Hymenoptera: Chalcidoidea) with a key to genera. Ann.
Carnegie Mus. 44: 171±181.
Werren, J.H. 1983. Sex ratio evolution under local mate
competition in a parasitic wasp. Evolution 37: 116±124.
Whiting, A.R. 1967. The biology of the parasitic wasp Mormoniella vitripennis [ ˆ Nasonia brevicornis] (Walker). Quart. Rev. Biol.
42: 333±406.
Received 20 July 1998; accepted 23 July 1998
J. EVOL. BIOL. 12 (1999) 586±595 ã 1999 BLACKWELL SCIENCE LTD