Understanding the evolution of sexually dimorphic development is a key goal in many biological contexts [1
]. However, in many taxa the ability to investigate dimorphism in early gene expression and development is impeded by the lack of a reliable method for assigning sex to embryos, larvae or other early life stages. For example, in holometabolous insects adult morphology is determined largely prior to formation of the primary sex organs and eclosion. Investigation of sexually dimorphic development would be greatly facilitated by the identification of sex in the absence of traditional cues such as gonads or genitalia.
Hypercephaly, in the form of lateral extensions of the head capsule, is observed in several families of Diptera [4
]. A particularly exaggerated form is found in the Diopsidae (stalk-eyed flies) in which eyes and antennae are laterally displaced at the end of eyestalks in both sexes. In many Diopsid species eyespan (the distance between the eyes) is sexually dimorphic, with males being larger than females. Eyespan and the degree of sexual dimorphism in eyespan varies considerably within the family [5
], and there is empirical evidence that the highly exaggerated eyestalks found in males have evolved under sexual selection through strong female mate preference for males with larger eyespan [7
]. Eyespan is determined prior to eclosion and is sensitive to external stress during pupation (e.g. heat shock; [11
]). Developmental studies indicate that the expression of key regulatory genes involved in early head capsule specification is similar to that observed in other Dipterans [12
From an evolutionary standpoint there is considerable interest in identifying the timing of expression of the hitherto unknown genes on which selection acts to regulate eyespan in sexually dimorphic Diopsid species. One potentially powerful approach is to compare gene expression during the development of eyestalks in between the sexes, since sexual dimorphism likely results from differential gene expression with respect to sex. However, this requires a method for determining the sex of pre-adult flies. In the model Dipteran Drosophila melanogaster it is possible to do this either on the basis of gonad size (the male gonad being significantly larger than the female in 3rd instar larvae), or by using X-linked genetic markers such as the cuticle pigmentation gene yellow. Neither method is applicable in stalk-eyed flies in which the larval gonads of both males and females are equally small and undifferentiated and visible genetic markers are lacking. Here we present an alternative, novel method based on genital disc morphology and X-linked DNA markers, which can be applied in principle to many different Diopsid species, and potentially to other taxa.
The adult structures of Dipteran flies, including stalk-eyed flies, develop from the larval imaginal discs, which originate as invaginations of the embryonic ectoderm. In D. melanogaster
it has been shown that with the exception of the eye-antennal and genital discs all of the imaginal discs are composed of cells derived from a single embryonic segment [15
]. Imaginal disc cells multiply throughout larval development and differentiate during metamorphosis. While the other imaginal discs exist as pairs, in Drosophila
the genital disc is a single disc in each sex [16
the genital disc develops into the internal and external genitalia and the analia, collectively known as the terminalia. It is the only imaginal disc with an unambiguous sexually dimorphic morphology. The genital discs of both sexes comprise cells from the eighth, ninth and tenth/eleventh abdominal embryonic segments and the fate of these cells depends on the fly's sex [17
]. In male larvae, the cells of the eighth segment have repressed proliferation, as these cells would otherwise develop into the female genitalia; while in female larvae, cells from the ninth segment, which would otherwise give rise to the male genitalia, do not develop. In both sexes cells from the tenth/eleventh abdominal segments develop into the analia [19
In the model stalk-eyed fly species, Teleopsis dalmanni
(previously known as Cyrtodiopsis dalmanni
]), we found that the genital discs have two distinct morphologies. We have termed these "compact" and "lobed" morphologies to reflect the fact that the former morph is relatively flat and compact. In D. melanogaster
, segment polarity genes such as engrailed
) have very different expression patterns in the male and female discs, which reflect later differences in eventual segment fate [21
]. The pattern of en
protein expression in T. dalmanni
genital discs showing the compact morphology resembled that observed in the male genital discs of D. melanogaster
. As in D. melanogaster
, male Diopsids have a single X
chromosome whilst females have a pair [22
]. We confirmed the proposed association between disc morphology and sex by evaluating the combined heterozygosity of four X
-linked microsatellite markers; individuals with a compact form of genital disc had hemizygous (male) genotypes, whereas individuals with lobed discs tended to exhibit heterozygous (female) genotypes.
To determine the extent to which male and female genital disc structure and segmental composition is conserved across the Diopsid clade, we characterised morphology and EN expression in a panel of Diopsid species spanning three genera. A compact/lobed dimorphism similar to that of T. dalmanni was observed and in all species studied EN expression patterns were consistent with the compact morphology being typical of males across genera.
We exploited our ability to assign sex by genital disc morphology to compare cell division rates in male and female eye-antennal discs. In T. dalmanni the larger eyespan of males is due, at least in part, to the eyestalks being composed of more cells. This could reflect a higher rate of cell division or a prolonged period of proliferation in males. We used genital disc morphology to determine whether the frequency of cell division differs between male and female H3 antibody-labelled eye-antennal discs in two sexually dimorphic species, T. dalmanni and Diasemopsis meigenni. We found no significant differences in dividing cell counts between the sexes in either species.