Variation of sperm length and heteromorphism in drosophilid species

La longueur des spermatozoides mesures chez 27 especes de Drosophilides est etudiee au niveau de l'ensemble de la famille (75 especes)


INTRODUCTION
In most Eucaryote species, meiotic reproduction has evolved in producing two sizes of gametes, known as the macro, or female gamete (oocyte or ovum) and the micro, or male gamete (sperm) respectively (Parker et al., 1972;Power, 1976;Maynard Smith, 1978;Alexander & Borgia, 1979;Parker, 1984). In the macrogamete, size variations between taxa are well documented and incorporated in the evolutionary theories of parental investment (Trivers, 1972) and life history strategies (Throckmorton, 1966).
By comparison, evolutionary trends in the microgamete (sperm) have remained neglected. Indeed, when considering vertebrates an overall uniformity seems the rule, yet some differences in sperm heads or tails have been found in rabbits and rodents (Friend, 1936;Beatty & Napier, 1960;Beatty & Sharma, 1960;Woolley, 1971). In these, sperm with a medium size flagellum less than 0.1 mm long are produced in huge numbers and each gamete has an extremely low probability of producing a zygote. Sperm shape, size and ultrastructure are, however, far more diverse among invertebrates (Fain-Maurel, 1966;Afzelius et al., 1976;Baccetti, 1979;Sivinski, 1984;Chauvin et al., 1988) and analysing this diversity should help to understand the developmental constraints and selective pressures which permitted or promoted the various patterns presently observed. In some invertebrates, including Lepidoptera, two or more morphologically and functionally different microgametes occur intraspecifically and warrant recognition of eusperm and parasperm (Healy & Jamieson, 1981;Jamieson, 1987a).
In the present work, attention is focused on the evolution of sperm length in a monophyletic dipteran taxon, the family Drosophilidae. Drosophila melanogaster, generally considered as a reference for this group, is known for its long sperm (1.9 mm) which is almost as long as the body of the fly (Cooper, 1950;Yanders & Perras, 1960;Beatty & Burgoyne, 1971;Gould-Somerot et al., 1974;Joly, 1987). However, at the family level, D. melanogaster sperm can appear very short compared to those of other species such as for instance D. hydei, where it can be 4-5 times (1.4 or even 1.9 cm) the size of the body of the fly (Hess & Meyer, 1963;Jamieson, 1987b).
Taxonomists have long been aware that, in the family Drosophilidae, sperm length could be very variable between species and several papers have been recently devoted to this problem (Beatty & Sidhu, 1970;Sanger & Miller, 1973;Gromko et al., 1984;Sivinski, 1984;Hatsumi & Wakahama, 1986;Hihara & Kurokawa, 1987;Joly, 1987). Explaining such variations raises an evolutionary challenge which may be formulated as follows: if sperm length is a fast evolving trait, it should exhibit a high genetic variance and a high heritability, at least in some species; moreover, a rapid evolution for increased or decreased length would be difficult to explain if the trait is considered as neutral, and strong selective pressures should exist or have existed during the process of speciation.
In Drosophila, intraspecific genetic variability of sperm length is poorly documented and presently available investigations have failed to demonstrate genetic variance (Joly, 1987) or have found only very limited variability (Beatty & Sidhu, 1970;Sanger & Miller, 1973). The aim of this paper is to focus attention on interspecific variations and to present an overview of what is known in the family.
In the genus Drosophila, sperm length can indeed be considered as a fast evolving trait with a loose relationship with phylogeny. In most species, sperm length distributions within the individual are unimodal with a limited variability. However, in one monophyletic taxon, the D. occurs species group, the occurrence of bimodal distributions seems the rule and we therefore argue here that it corresponds to an evolutionary stable strategy (ESS) (Maynard Smith, 1974).

MATERIAL AND METHODS
Sperm length was measured in 27 species, 13 of which belong to the obscura species group including the two recently discovered East African species (D. microlabis and D. kitumensis) (Cariou et al., 1988). The source of material of the obscura group species investigated was the same as in Cariou et al. (1988) with the exception of D. a f finis (14012, 014-1) and D. azteca (14012-0171) which were provided by the Bowling Green Stock.
The eight species of the melanogaster subgroup were analyzed. The source of specimens of the melanogaster complex species was the same as in Joly (1987) while those of the others were the following: D. teissieri and D. yaku.ba came from different localities in Africa (Gif Stock); D. erecta (Ivory Coast, Gif 220-5) and D. orena (West Cameroon, Gif 188-1).
The strains were reared at 21 °C. Sperm were recovered from the seminal vesicles of one or several males. The testes were isolated and opened in a drop of saline solution and the sperm allowed to spread out. This preparation was observed under a microscope with phase contrast optics. When the sperm had ceased to move, they were traced with the aid of a camera lucida and the trace lengths measured with a cursor on a digitizing table connected to a microcomputer. Except for the obscura group species, the measure of cyst length was preferred to that of sperm length to minimise the risk of breakage. All details of this method are given in Joly (1987 and1989).
The sperm length of the 48 other species belonging to different taxa of the Drosophilidae are provided in the literature (Sanger & Miller, 1973;Hatsumi & Wakahama, 1986;Hihara & Kurokawa, 1987).

RESULTS
Results for the investigated species are given in Table I and for the 13 species of the D. obscura group in Table III. Some of the species presented in these tables have already been studied by other investigators, for example D. melanogaster (Table IV) and some species in the D. obscura group. Our measurements are, on the whole, in good agreement with previous data in spite of methodological problems mainly due to the difficulty in obtaining identifiable and unbroken cells. It is therefore possible to present (Table II)  At a genus level, mean length varies from 0.63 (Amiota) to 5.32 mm (Mycodrosophila). However, among the 75 species presently studied, 64 belong to the Drosophila genus which is itself characterized by a huge interspecific heterogeneity. A more detailed analysis according to taxonomic subdivisions is presented in the lower part of Table II. The best documented subgenera, Drosophila and Sophophora, exhibit significant sperm length variations, with means of 5.03 and 1.14 mm respectively. Also, within each subgenus, lower taxa, i.e. species groups, may have different lengths and variations. For example, in Drosophila, flies in the D. immigmns group have much shorter sperm than in both the D. repleta and D. virilis groups which are characterized by very long sperm; the record length is provided here by D. littoralis from the latter group where it reaches 2 cm, that is 6 times the body length (Table I). In Sophophora, we may further contrast the D. medanogaster and the D. obscura species groups with means of 1.45 and 0.30 mm respectively.
Another way of analysing the data is to consider the heterogeneity among species belonging to taxa of similar levels. Since mean lengths are so variable, variances cannot be used directly and a relative measure, the coefficient of variation (c.v.) has therefore been preferred. This analysis is limited to Drosophila, since other genera are poorly documented. On the other hand, the Drosophila genus comprises so many species (over 1 500) that taxonomists felt the need for a series of hierarchical subdivisions, as defined, for example, by Bock & Wheeler (1972) who recognized subgenera, species groups, species subgroups, and within the latter, species complexes, species &dquo;clusters&dquo; , and pairs or groups of sibling species.
At the genus level, the overall c.v. is 130% (Table II) which means that the standard deviation is higher than the mean and the actual distribution is strongly skewed towards high values. Considering lower level taxa leads to lower values of c.v., i.e. 96%, 61% and 34% respectively for subgenera, species groups and species subgroups. It appears that homogeneity increases when more closely related species are compared. It has been shown that intraspecific genetic variability in sperm length is poorly documented in Drosophila and requires further investigation. However, within each species, the shape of the distributions of individual sperm measurements is worthy of consideration, and examples of such distributions are given in Figure 1. In D. melanogaster and D. simulans the distributions are obviously unimodal and close to a gaussian curve. Such is not the case in species of the D. obscura group, which exhibit clear-cut disjoint distributions. This intraspecific and intraindividual heterogeneity was already known in some of these species and the word polymegaly, meaning several sizes, was coined to describe this situation (Beatty & Sidhu, 1970;Beatty & Burgoyne, 1971). Our results confirm and extend these observations. In some cases, such as D. pseudoobscura, it could be argued that, by visual inspection, several peaks may be recognized. However, no statistical method exists for counting the number of peaks in a distribution. On the other hand, visual inspection always shows a well defined peak for short sperm while the situation may be more complex for longer sperm. As a conservative measure, it was decided to differentiate only two size classes in each species, i.e. short and long sperm, the size limit between the two classes being in most cases easy to define. Morphometric data, analysed in this way, are presented in Table III for the 13 investigated species of the D obscura group. size of mammals. The interspecific variation ranges between 0.056 and 0.143 mm. By contrast, the long sperm class is more variable, ranging from 0.139 to 0.925 mm, and is also more heterogenous, as shown by its high c.v. When the two classes are pooled, the bimodality of the distributions is evidenced by the very high c.v. : 53%.

DISCUSSION AND CONCLUSION
The great length of the sperm of numerous drosophilid species raises some technical problems concerning length determination: very elongated flagella are easily broken during dissection and, taking into account incomplete cells, would both decrease the calculated mean and increase the variance.
For that reason, measurement of mature cysts, assumed to give more reliable data, was preferred in our study for species with longer sperm, e.g. in D. !n,elanogaster. This method in addition to the use of saline solution instead of fixatives, probably explains the discrepancies between our data and some of those previously published (Table IV). For extreme lengths of over one centimeter, found for example in D. littoralis.and D. hydei, even the cysts are often broken so that it is very difficult to evaluate intraspecific variability. However, it seems reasonable to conclude that shorter values correspond to incomplete cysts and to consider only the longer measurements as typical of the species.
In contrast, there are no technical difficulties in having complete short sperm which do not break easily. Therefore, the heteromorphism of the distributions in the D. obscura group species, which has already been observed by previous investigators (Yanders & Perras, 1960;Beatty & Sidhu, 1970;Policansky, 1970;Beatty & Burgoyne, 1971;Sanger & Miller, 1973;Kurokawa et al., 1974) cannot be accounted for by any technical bias.
The occurrence of very long male gametes in numerous Drosophila species raises several evolutionary questions, to be discussed below. The first concerns the ancestral or primitive state of sperm length. According to theories of modern cladistic systematics, this may be inferred by considering taxonomic outgroups. There are very few cases of animals with such relatively giant sperm. Among these are featherwing beetles (Coleoptera, Ptiliidae) (Dybas & Dybas, 1981;Taylor, 1982) or some ostracods (Bauer, 1940) where sperm may be several times the male length (see review in Sivinski, 1984;Jamieson, 1987b). Nevertheless, species with sperm of inordinate length are still more common in fruit flies. A reasonable proposal is therefore that short sperm are primitive while long sperm are derived (Hihara & Kurokawa, 1987). However, the situation is less clear at a lower level; in the D. melanogaster species complex, for instance (Table I), sperm length distributions appear to be divergent in most closely related species (e.g. D. simulans cf. D. sechellia), but convergent in less closely related species (e.g. D. secheldia cf. D. melanogaster). Here if phylogeny is considered (Cariou, 1987) we must conclude that elongation occurred independently during the speciation process. A similar reasoning could be applied in comparing other taxa in which there are species with either short or long sperm. Unfortunately, knowledge of Drosophila phylogenies does not presently allow such comparison. Whatever the conclusions might be, it remains clear that evolution and speciation in the family Drosophilidae is characterized by a general tendency towards increasing sperm length, as already assumed by Hihara & Kurokawa (1987).
This overall evolutionary tendency further suggests that size variation is not random but has been subject to natural selection (Joly, 1989). The most important questions that then arise are: how and why did sperm elongation evolve? Some insights may be gained by considering the heteromorphism of the D. obscura group species.
Clearly, heteromorphism, which is typical of the whole group, is genetically determined (Beatty & Sidhu, 1970). Moreover, this does not correspond to a genetic polymorphism at the diploid level, since any single male produces heteromorphic sperm. Nor is it a case of gametic polymorphism since all the sperm cells, included in the same cysts, and which could be genetically different, exhibit the same length. Heteromorphism seems to be more a case of polyphenotypism which is determined by some unknown physiological mechanisms at the cyst level. A reasonable interpretation is that heteromorphism is an evolutionary stable strategy (ESS) (Maynard Smith, 1974), each sperm class having some adaptive advantage.
For instance, the short and long sperm may have differential capacities both to reach the storage organs (preemption capacity) and to resist the second male paragonial substances (antipreemption capacity) when a female remates. A precise formulation of such a hypothesis, which requires comparison of the evolution of both sperm length and mating systems in different species, is proposed in a forthcoming paper.